Raising Agricultural Yields Spares Land

Farmland expansion is one of the leading drivers of biodiversity loss and greenhouse gas emissions globally. It is not surprising, then, that the question of how to slow or halt the growth in agricultural area has been at the top of the agenda in debates around agriculture. With a global population growing larger and wealthier by the day, finding ways to reduce farmland expansion without relying solely on rich countries moderating their food consumption is essential.

Much of this debate is framed around sustainable intensification—the idea that higher productivity and lower environmental impacts can meet the twin challenges of food security and environmental protection. A fundamental premise of this idea is that higher productivity, and specifically higher yields, will lead to less land being used for crops and pastures—a phenomenon known as land sparing.

However, a growing number of voices have argued that higher yields may spare very little land from conversion or even increase land use, an outcome termed "backfire," also referred to as Jevons Paradox. "Land-sparing is a weak process that occurs under a limited set of circumstances," according to Robert Ewers and colleagues. Andrew Kniss and others say that "yield gains have not been clearly linked with increased land set aside for conservation at the global or regional scale." Claire Kremen and Adina Merenlender argue that "it is a fallacy that such [simplified and intensified production systems] will ultimately spare more land for nature conservation."

But the situation is not as bleak as one might be led to believe by these and many other authors. Locally, land sparing is not guaranteed, but it occurs under many circumstances. And globally, despite reports to the contrary, land sparing is all but guaranteed and likely of very substantial magnitude.

It's hard to overstate how important it is that we get this right. The claims that land sparing is minimal or that backfire is the norm imply that investments in yield-boosting technologies and practices have far fewer benefits than would otherwise be the case. If higher yields don’t lead to more land for conservation and, by extension, to more carbon stored in ecosystems, there is less reason to implement more intensive farming systems that are often seen as causing negative side effects. Better, in that case, to de-emphasize yields and shift investments to other areas.

However, yield improvements really do lead to land sparing (if not locally, then globally). Claims to the contrary could lead to a misallocation of investments and, consequently, worse environmental (and social) outcomes. Facing ever-increasing food demand and a pressing need to reduce carbon emissions, the results of a false narrative that land sparing is not real could be catastrophic.

From backfire to land sparing

When a farmer adopts a more productive technology—or when Total Factor Productivity, as it is referred to in the literature, goes up—by definition it means she'll be able to grow more crops with the same amount of inputs. If land is one of the inputs (or “factors”) with higher productivity, less land is needed to produce a given amount of crops. This efficiency gain is where land sparing comes from.

At the same time, higher productivity means it costs less to produce the same amount of food. This has two effects.

First, since crops are now cheaper, consumers will demand more, which—all else equal—will increase the area of cropland. How much so depends on how responsive consumers are to prices. Demand for staple crops tends to be fairly stable since people don’t increase their consumption much even if prices go down. Therefore, land sparing is more likely to take place, even at the local level, when yields of staples go up. The situation is different for some internationally traded products like soy and palm oil. If farmers across South America get better at producing soy, their large market share and the high demand response to the lower prices means the effect is quite likely to be an increase in cropland regionally.

Second, the farmer can sell her produce at a lower cost than before and, as such, gains a competitive edge, which incentivizes her to expand production. This rebound will almost always offset some of the land sparing from increased yields. In some cases, the rebound is so strong that we are left with more land in production than would have been the case without the productivity improvement—that is, backfire occurs.

If we stop here, as some studies essentially do, it's not far-fetched to conclude that yield improvements are likely to cause more land to be taken into production. This is also the expected outcome from a widely used model of agricultural production first developed by Johann Heinrich Von Thünen in the 1800s. There are a fair number of local case studies that support this idea of backfire, some of which are contained in a 2001 volume edited by Arild Angelsen and David Kaimowitz. In their introduction, the two of them concluded that "trade-offs and win–lose between forest conservation and technological progress in agriculture in areas near forests appear to be the rule rather than the exception."

But the story doesn't end there. When producers in one area increase their supply, prices go down. For the farmers responsible for the productivity improvement, this price effect counteracts the increased profitability from higher productivity, thereby dampening rebound. For farmers elsewhere, the lower prices lead to lower production and, thus, less land used for farming. The net effect of yield improvements in one place can therefore only be assessed by adding up the change in cropland area in that region and in the rest of the world. This is the part of the story that many who downplay land sparing most often miss.

The degree to which prices go down—lowering profits from farming—depends on various factors. Rebound might be small if the productivity improvement takes place for a large share of farmers in a given market. For example, if products are mostly sold in a local market, the price-suppressing effect is likely to be large enough to discourage farmers from expanding the area under production. The same thing happens if most farmers across the world benefit from the productivity improvement. This price effect is also mediated, especially at the local level, by institutional, technical, economic, and labor-related factors.

Where does this leave us? At the local or regional level—that is, the area where the yield improvement took place—both land sparing and backfire are possible. It is impossible to say just how common each of these outcomes is, but there are a lot of local studies (such as this one by Ryan Abman and Conor Carney) that find evidence of land sparing. In a more recent review than that of Angelsen and Kaimowitz, Nelson Villoria, Derek Byerlee, and James Stevenson conclude that "the empirical support of Jevons’ paradox (i.e., technological progress leads to more deforestation) is much weaker than what seems to be accepted in the literature." At the global level—that is, when we also account for land-use changes outside the area where the yield improvement took place—land sparing is virtually guaranteed, as we will see below. In other words, land sparing almost always occurs—it is just a question of where.

The evidence for global land sparing

To see how land sparing plays out globally, we can look to so-called equilibrium models that describe the relationships between agricultural production, land use, demand, and so on, in response to an improvement in productivity in a given region. There are very few studies that do this, the most prominent ones being written by James Stevenson and colleagues, and by Thomas Hertel and others. These two papers ask how much land was spared as a result of the Green Revolution, specifically the productivity improvements in certain developing countries stemming from research and diffusion of improved seeds by the Consultative Group for International Agricultural Research, or CGIAR.

The studies compare a scenario replicating historical trends in overall crop productivity to a counterfactual, lower-productivity scenario where the improved seeds were not adopted in the developing world and population, income, and food demand were thus different. The difference in total cropland between these two scenarios is the land sparing attributable to the better seeds.

Their results show that the improved technology in the Green Revolution region (Asia, Latin America, and the Middle East) resulted in higher land use within the region, but a net land sparing at the global level. Their estimates of the magnitude of this effect are quite different: Stevenson et al. put it as 17.9 to 26.8 million hectares, whereas Hertel et al. find it to be 144 million hectares. This discrepancy comes down to how the models are built, but both estimates point to a land-sparing effect that is significant in terms of protecting biodiversity and lowering greenhouse gas emissions. Hertel et al.’s estimate suggests that productivity improvements reduced cropland expansion by over 50%, sparing an area almost half the size of Western Europe.

But even these estimates are likely below the full land-sparing effect of global yield improvements. There are several reasons for this: they only account for the part of productivity improvements that are attributable to seeds and thus exclude other factors like agronomy; they only look at the effect of productivity improvements in part of the developing world, thus omitting about half of global crop production; they only look at certain crops; and they do not include any spillover of innovations from the Green Revolution area to the rest of the world.

A different study, by Nelson Villoria, uses a different framework based on regressions and only looks at the period 1991-2010 but arrives at broadly consistent findings: productivity growth has often increased land use in the region where it occurs but spared land globally. Regional land expansion is particularly pronounced in countries highly integrated into global markets like Argentina, Australia, and Malaysia, whereas regional land sparing has occurred in more insulated countries in developing Asia and Sub-Saharan Africa, such as India and Nigeria.

Villoria estimates that without productivity improvements, an additional 173 million hectares of land would have been required, a number even higher than that of Hertel et al. but over a much shorter time period. If we extrapolate to the full period since 1960, we'll likely end up with several hundred million hectares. Another way to look at this result is to put it in the context of the cropland expansion that the world actually saw, which was about 167 million hectares between 1961 and 2010. Without yield improvements, therefore, we would have seen more than double the cropland expansion. Villoria's paper has the advantage of being more empirically based than the general equilibrium models employed by Stevenson and Hertel, and Hertel himself told me that he considers Villoria's results "pretty definitive."

Net global land sparing, while the norm historically, is not guaranteed. Thomas Hertel and others present a future scenario where the opposite would occur if Sub-Saharan Africa saw its agricultural productivity go up along the lines of what happened in Asia and elsewhere during the Green Revolution. If historical trade patterns persist, a Green Revolution in Sub-Saharan Africa would be globally land-sparing. However, if one assumes completely integrated global commodity markets—where trade flows freely across borders—the effect would be increased global land use. This is not a realistic prospect in the foreseeable future, but still useful for illustrative purposes.

In this case, we would have a sort of perfect storm in terms of the factors that make backfire more likely. In particular, we would have a large, global market, where Sub-Saharan Africa is still a relatively small player, so it could substantially increase its global market share. Furthermore, cropland area in the region is particularly responsive to demand increases. What really tips the scales, though, is that Sub-Saharan Africa's yields are currently far below the global average. This means that, as the region takes on a larger share of global production, the average global yields would drop significantly.

Yet even in this extreme case of perfect global markets, an African Green Revolution would eventually spare land. As the productivity improvements take hold, the gap between Africa's yields and those of the rest of the world would close, such that this region’s increased market share no longer depresses global average yields. Hertel et al. estimate that the crossover point, after which further productivity growth would be land-sparing, would happen after about 20 years of sustained productivity growth.

The double-edged sword

So far I've focused on the effects of improved technology on land use, finding that global land sparing is by far the most likely outcome, but that local land sparing does not always take place. This, however, might not be a very satisfying situation for policymakers who want to ensure that more land can remain as natural habitat in the same place that the agricultural intervention took place. This dilemma has been described as a "double-edged sword" by L. Roman Carrasco and others. The authors gain this insight from a study showing that improved yields in Indonesian and Malaysian palm oil would spare 300,000 hectares of land globally—much of it in Brazil, India, and Canada—but lead to the loss of 112,000 hectares of tropical forests in Malaysia and Indonesia. With biodiversity and carbon stocks often being much higher in the tropics than in temperate regions, the net environmental effect of this reshuffling of land use is far from clear.

The good news here is that there is a whole suite of tools that can mitigate local farmland expansion. Ben Phalan and others grouped these tools into four categories. First, land-use zoning, such as protected areas, can help ensure that critical habitats are kept from agricultural encroachment. The second category includes economic instruments like payments for forest protection, land taxes, and targeted subsidies. Payments have proven successful in some cases, such as in a randomized study in Uganda. Land taxes could discourage deforestation but are rarely used in practice.

The third set of instruments falls under the banner of what Phalan et al. refer to as “spatially strategic deployment of technology, infrastructure, or agronomic knowledge.” One example of this comes from India, where a road-building program was shown to increase tree cover in more densely populated areas. Finally, Phalan et al. point to standards and certification, which include efforts like the Amazon Soy Moratorium and the Roundtable on Sustainable Palm Oil. Robert Heilmayr and colleagues showed that the former spared on the order of 18,000 square kilometers of forest in Brazil.

All of these have the potential to prevent backfire and can therefore ensure that forests and other natural habitats are indeed preserved in the places where policymakers have encouraged greater agricultural yields. But they all come with indirect effects that can be either positive or negative for conservation. On the positive side, interventions like protected areas can make it more attractive to invest in capital rather than forest clearing, thereby leading to higher crop yields, which in turn have a land-sparing effect (somewhere!).

On the other hand, any attempt to reduce agricultural expansion and thus agricultural production in one place is likely to lead to leakage, thereby increasing pressure on natural habitats somewhere else. When reductions in production—sometimes policy-driven—take place in regions with efficient agriculture and relatively low carbon costs, this may incur a cost in other places large enough to make for a net loss globally in terms of both biodiversity and climate change.

For instance, consider the US, where crop production often takes place on land that would not otherwise have been forested, making the carbon and biodiversity costs relatively low. From a global perspective, it might be better to intensify agriculture in the US and allow cropland expansion to occur domestically in order to spare land in regions with higher carbon stocks, such as South America.

Given America's disproportionate influence on global crop markets, improved productivity here has vast land-sparing effects in the rest of the world. Nelson Villoria estimates that for every 1% gain in Total Factor Productivity (inputs required per unit of output) in the US, croplands in the rest of the world shrink by 0.7%. Similarly, a Breakthrough Institute analysis estimates that the crop productivity growth from doubling public agricultural R&D spending in the US would increase US cropland area by about 1.8 million hectares but spare a total of 18.2 million hectares globally and reduce emissions by 109 million metric tons of CO2e per year. Perhaps the US and other similarly positioned regions should take one for the team and produce as much as they can?

Amidst these environmental arguments, we should not forget that many developing countries rely heavily on agriculture for their livelihoods. If the US boosts yields and absorbs more of global food demand, it means there’s less demand elsewhere in the world, potentially hurting farmers. There is no simple win-win. Developed countries should therefore consider pairing their own investments in productivity with support for similar efforts in the rest of the world.

Not just land

Another dilemma arising from the question of land sparing is that in the drive to improve yields and spare land, other forms of environmental harm might occur due to the increased use of inputs like fertilizers, pesticides, or irrigation. It's important to note that in much of the world in the last few decades, increases in output have to a large degree come from pure productivity gains rather than higher inputs, but in developing countries, increasing inputs is still an important part of agricultural intensification. Countless studies have documented how agricultural intensification can reduce biodiversity and increase pollution.

But just like we've seen before, a local-only perspective can be myopic. In many, if not most cases, intensive agriculture has lower emissions per unit of output. So even though land-sparing intensification can lead to more pollution locally, it may very well reduce pollution at larger scales. In these cases, land sparing must be weighed against other environmental impacts—a decision that falls to local decision-makers and stakeholders. Inevitably, there will be winners and losers.

Don’t give up on yields

In sum, land sparing is common at the local level and the norm at the global scale, in contrast to recent appeals to Jevons' paradox. Therefore, yield improvements—as part of sustainable intensification—remain an essential tool in efforts to reduce the environmental impacts of agriculture, especially when thinking globally. Innovations in seeds and agronomy; streamlined regulation to allow for the diffusion of those technologies; agricultural extension to bring better tools and practices to farmers – these are only some of the practical ways in which higher yields can be achieved, and they all need heavily ramped-up investment by the private and public sector, lest yields fall behind demand and cropland continue expanding.

Yield-boosting measures, however, always need to be thoughtfully paired with other measures to ensure that outcomes—environmental as well as social—are acceptable at the local level. Phalan et al.'s four-part recipe is a good way to start. This can, where appropriate, help bring about the environmental benefits of land sparing to the place where intensification occurred.

Farmers' and local communities' welfare must always be considered; environmental protection must never be done on their backs. Because here’s the thing: rebound and backfire can be a good thing for local farmers. The very mechanism for rebound is that local farmers become more profitable and see higher demand for their products; that’s why they may choose to expand their farms. Holding farmers back from exploiting this increased productivity without fair compensation is not a viable option. Protected areas and other measures must be anchored with local communities and not created at their expense.

What this also tells us is that, at the end of the day, the argument for boosting yields—especially among smallholder farmers—is not just environmental: it is humanitarian. As such, it's not really a question of whether we should pursue productivity improvements or not, but how we do so. The farming of the future does not—and probably should not—look like the “conventional” farming we have in the US and elsewhere today. Environmental impacts must be drastically cut by using smarter and more efficient technologies and practices and by integrating agroecological methods where that makes sense. But the focus on yield as one of the most important priorities should not be scrapped.