Similarities in USDA’s International Baseline Projections and their Relationship with Projection Accuracy

by Rabail Chandio and Ani Katchova

The US Department of Agriculture (USDA) releases annual international baseline projections that play a key role in shaping farm policy, guiding market outlooks, and informing stakeholders about how supply, demand, and trade might evolve over the next decade. Just as farmland value surveys or commodity price outlooks assist local producers in Iowa and the broader Corn Belt, these USDA baselines are critical for understanding broader trends in US agriculture. However, compared to short-term forecasts—like monthly crop production estimates—these long-term projections have typically received less attention.

In our recent article in Journal of Agricultural and Applied Economics Association (Chandio and Katchova 2024), we investigate an under-explored aspect of these international baselines: the extent to which forecasts for different countries might converge—often toward major producers like the United States—and whether these similarities are related to the accuracy of the forecast. Below, we share our approach and key insights that may be of interest to producers, agribusinesses, and policymakers who rely on these baselines for strategic guidance.

Why focus on similarities? 

USDA’s international baseline projections are created through a collaborative process where model-based outputs are adjusted based on expert judgment. This collaboration, while necessary, can lead to some countries’ forecasts resembling those of more data-rich or influential regions—most often the United States. In situations where robust local data are limited, experts may lean heavily on the United States’ path to fill in the gaps. 


Such a strategy can work well if the United States truly serves as a good reference point. At the same time, it might fall short if local conditions—from climate and labor availability to policy and market demand—differ significantly. In our work, we center on three major commodities—corn, soybeans, and wheat—which together form a large portion of global grain and oilseed trade. Within each crop, we examine yield, harvested area, imports, exports, total consumption, and ending stocks—the core set of supply and demand indicators that shape how markets function worldwide.

Does similarity help or hinder accuracy? 

To assess whether these similarities matter for accuracy, we measured each country’s projection error by comparing the baseline forecasts to the actual data once it became available. We then examined if a closer resemblance to the United States reduces or increases projection errors. We replicated this process using China and Brazil as alternate “base countries,” acknowledging both nations’ growing role in global grain and oilseed markets. 


Note that our analysis focuses on correlations: a strong similarity might reflect “herding,” where forecasters align with a dominant viewpoint. Whether that helps or hurts varies by commodity and variable.

Key findings

  1. Yield forecasts are most similar across countries: We discovered that yields, especially for corn, tend to have the smallest projection distances—meaning countries’ projections often look very much alike. Yields typically shift more slowly over time, so a single assumed annual productivity growth could plausibly get extended across multiple regions. However, when we checked actual yield differences, they were sometimes larger than the projected differences. This suggests there may be more alignment on paper than local agronomic realities would justify.
  2. Harvested area and ending stocks diverge more: In contrast, we found that harvested area and ending stocks showed much higher projection distances across countries. These variables can be heavily influenced by domestic policies, local land constraints, or strategic storage decisions that differ from US norms. When we see large distances in the forecasts, it may reflect truly distinct assumptions or deeper uncertainty about local conditions.
  3. When similarity helps—and when it hurts: We looked at how similarity correlated with accuracy. In some cases, a country that closely mirrored the United States ended up reducing its forecast errors. Examples include, corn yield, corn harvested area, corn exports, and wheat imports. For these, aligning a smaller or data-deficient country’s projections with the more robust US outlook seemed beneficial. However, for other variables—particularly soybean imports, wheat harvested area, and total consumption—being too close to the US forecast was associated with larger errors. It appears that demand-related variables are more localized and require region-specific information.
  4. China or Brazil as an alternative benchmark: Recognizing that China and Brazil are also big players, we replicated the analysis using them as the base country. In some instances, aligning with Brazil’s corn outlook resulted in stronger accuracy gains than aligning with the United States. Similarly, China could be a better anchor for soybean imports, given its enormous share of global soybean demand. Our takeaway is that “one-size-fits-all” anchoring to the United States is not necessarily the best strategy for every commodity and variable—nor is ignoring the United States entirely.
  5. Projection horizon matters: Forecast accuracy typically worsens at longer horizons (e.g., 7–10 years; see figures 1 and 2). Our study confirms this, but we also found that the effect of “herding” or “diverging” can magnify over time. If a region’s path is overly synchronized with a major producer’s assumptions, any local differences that surface down the line can lead to large forecast misses.
Figure 1. Projection error for corn yield by projection horizon.
Figure 1. Projection error for corn yield by projection horizon. 
Note: USDA’s baseline projections extend up to 10 years into the future. This figure illustrates how, for the countries shown, the projection error in corn yield shifts as the forecast horizon increases.
Figure 2. Relationship between projection error and projection distance from United States.
Figure 2. Relationship between projection error and projection distance from United States. 
Note: This figure shows how much the projection error (in percentage terms) changes when the distance between other countries’ USDA forecast and the US forecast increases by 1%. The bars above and below each estimate are the standard errors, which indicate the precision of those estimates.

From Farm Bill debates to international trade negotiations, USDA baselines inform a wide range of policy decisions. Recognizing that certain country forecasts may sometimes systematically track US assumptions gives policymakers a more nuanced understanding of these projections and helps ensure more informed, effective usage. Our analysis shows that USDA international baselines can show significant similarity across countries’ forecasts—especially for yield—while other variables reveal bigger divergences. Importantly, whether this similarity helps or harms accuracy varies by commodity and measure. For instance, aligning yield and harvested area forecasts with a big global producer’s outlook might be beneficial, but clustering around US assumptions for imports or consumption sometimes increases errors.

References 

Chandio, R., and A.L. Katchova. 2024. "Similarities in the USDA international baseline projections." Journal of the Agricultural and Applied Economics Association. 3: 505–21. https://doi.org/10.1002/jaa2.129.

Suggested citation

Chandio, R., and A. Katchova. 2025. "Similarities in USDA’s International Baseline Projections and their Relationship with Projection Accuracy." Agricultural Policy Review, Winter 2025. Center for Agricultural and Rural Development, Iowa State University. https://agpolicyreview.card.iastate.edu/similarities-usdas-international-baseline-projections-and-their-relationship-projection-accuracy.