Global Crop Yield Forecast Contracts

Of all the potential effects of climate change on people, its impact on agriculture and, thus, food security has arguably received the most attention. By 1995, sufficient research had accumulated to warrant an entire review chapter in the Intergovernmental Panel on Climate Change’s (IPCC) Second Assessment Report. Since then, research has continued to accumulate, and the IPCC has consistently emphasized the negative impacts of climate change on agriculture through the most recent Sixth Assessment Report.

These negative impacts have featured prominently in models of the economic impact of climate change since their inception, and the agriculture sector is still considered to be one of the largest contributors to the overall social cost of carbon. When those most concerned about climate change envision societal collapse, they often evoke the failure of the food system and famines as the precipitating event.

While many studies indicate an overall negative impact of climate change on food production, the situation is not straightforward. There are three primary ways in which CO₂-induced climate change affects crop productivity, and these do not represent universally detrimental effects.

The first is that there is a direct benefit of enhanced CO₂ concentrations. Increased atmospheric CO₂ levels can boost photosynthesis, resulting in greater plant growth and improved water-use efficiency. C3 crops, such as wheat, rice, and soybeans, can significantly benefit from additional CO₂, whereas C4 crops like corn do not, except during drought conditions. 

The second is that crops are affected by warming temperatures. Each crop, along with its varieties, has an optimal temperature for growth. As a result, crops currently grown below their optimal temperature may benefit from warming, while those grown at or above their optimal temperature, typically found in the tropics, will suffer due to rising temperatures. 

The third factor is that crops are affected by how much water is present. Just like temperature, crops require a specific level and distribution of water for optimal growth.If there is too little water, plants struggle to conduct photosynthesis; if there is too much, oxygen uptake by the roots is inhibited, thereby limiting growth. 

This all means that there is a tug-of-war of climate change effects on crops that will ‘net out’ differently depending on the crop, crop variety, and location. For instance, elevated CO₂ and climate change are projected to reduce corn yields across most latitudes and diminish wheat yields in tropical regions. In contrast, wheat, soybean, and rice are likely to benefit from CO₂  and climate change outside the tropics. 

From Our World in Data

Meanwhile, climate is far from the only driver of agricultural productivity. Over the past century, mechanization has replaced manual and animal labor, and the availability of synthetic fertilizers has enhanced soil nutrients. Additionally, the use of pesticides has controlled crop damage caused by insects. Breeding, both through conventional hybridization and, more recently, genetic modification, has provided plant strains with increased disease resistance, drought tolerance, and higher potential yields. Recently, ‘precision agriculture’ technology has integrated high-resolution satellite observations with weather forecasts, GPS-guided machinery, and real-time analytics to optimize inputs such as water, fertilizer, and pest control. 

In fact, the impact of various farming practices on crop yields has historically been far greater than that of climate change. For instance, while corn yields in Kenya have likely suffered due to climate change, this could be offset many times over if Kenyan farmers gain access to the resources available in high-income countries. 

From Our World in Data. 

Other key dynamics involve farmers adjusting their crop choices and planting schedules as temperatures rise. Additionally, markets and international trade can hypothetically facilitate the efficient redistribution of farm locations in response to climate change. 

The upshot is that the evolution of food production hinges on an ongoing interplay between climate change, technological innovation, the extent of mechanization, the adoption rate of advanced breeding techniques, farmer behavior, geopolitics, and international economics.

Given that the future of agriculture depends so much on drivers for which information is widely dispersed across disciplines, it is another quintessential question to pose to a prediction market. 

With this in mind, IBKR is introducing global average crop yield Forecast Contracts for 2030. There will be contracts for four crops—corn (maize), wheat, rice, and soybeans because together they account for nearly two-thirds of human caloric intake and have featured prominently in assessments of climate change impacts. 

Despite the complexities mentioned above, the global average yields of these four crops (mass of crop harvested per unit area of farmed land) have not displayed wild, unpredictable fluctuations. Instead, all of these crops have experienced relatively steady increases in yields over time, indicating that technological and economic advances have consistently outpaced any adverse effects of climate change.

Data from USDA FAS, plotted in Matlab.

Will these trends continue? To calculate the baseline expectation for yields in 2030, I essentially assume that the net effect of all the complex underlying drivers remains steady and that trends continue progressing as they have in the past. 

Technically, I use an Auto-Regressive Integrated Moving Average ARIMA(2,1,0) forecasting approach for these probabilities. The prediction intervals (and thus initial probabilities) in this ARIMA(2,1,0) framework are derived by taking into account the uncertainties in both the model parameters (the autoregressive coefficients and the overall variance of the historical residuals) and the accumulated step-by-step forecast errors. Conceptually, the model assumes that future differences in yield behave statistically like past differences, but uncertainty increases with time so the prediction interval widens as one projects further away from the last observed year. That structure alone would assume that whatever factors shaped past yield fluctuations will continue to manifest similarly in the future. This is, of course, not necessarily so, so to bluntly account for the Problem of Induction, I inflate the prediction interval by 25%. 

Participants in this market can judge for themselves whether this methodology constitutes a fair projection of future crop yields. 

Those who believe that technology and capital investment will grow faster than they have historically, or who expect a significant shift in seed breeding efforts, may consider the default yield projection to be too conservative, making the “Yes” side of the contract appealing (i.e., yields will exceed a particular threshold). Conversely, participants who are concerned about increasing climate impacts and stagnation in technological advancements may find the “No” side of the contract attractive.

Additionally, a farmer or anyone with a financial interest in crop productivity may find the “No” side of the contract beneficial as a hedging mechanism. If an investor is concerned that upcoming droughts, ongoing heatwaves, or adverse tipping points could jeopardize agricultural assets, purchasing “No” contracts could act as a form of insurance against these possibilities. 

Overall, these forecast contracts serve a dual purpose: they are instruments for participants to profit or hedge based on their views of future yields, and they also signal where collective expectations stand. For instance, if the “Yes” side becomes expensive because most participants believe yields will continue to rise, that indicates a general optimism about the balance of technological and economic progress despite any detrimental impacts from climate change. Alternatively, if the “No” sides gain momentum, it suggests significant market concerns that climate challenges and/or socio-political disruptions will impede progress.