Proposing valid, long term grain consumption scenarios – as required by players in the Australian grains industry – requires a different approach and tools to short-term forecasts, some of which are slightly fuzzy and challenging for us data driven analysts.

Sources of grain market information, such as price and volume data, are frequently accessed as they are often accurate, timely, thorough, simple to interpret, and easy to use. These traits make the information valuable to organisations who can quickly shift their focus to alternate markets at low cost. The information often usefully informs near-term forecasts.

However, for some players in the value chain, current market information, even information that projects a year out, may not be sufficient. For instance, cereal breeding companies need to accurately identify traits in new breeding lines that markets in the future will be willing to pay for. For a new variety to reach the market, breeding of the variety takes 5-6 years, and a lot of upfront investment, then another 2 or 3 years before the variety hits the market in commercial quantities. It is only once the variety is available in commercial quantities that the breeding company starts to recover its investment.

At any point in time, traders and marketers will prefer varieties with quality characteristics that markets are willing to pay for. If varieties do not deliver the quality wanted, they are downgraded and sold at a discount. This affects margins through the whole value chain and lessen farmers’ margins. So, if there is a 7-year gap between a breeder forming an opinion on markets’ future quality requirements and a subsequently bred variety being available in a market, then the breeder needs accurate information on what markets will prefer in the future.

A successful approach to developing future market scenarios requires more information than just trade history and price data. There is a need to know how demographics may drive future demand. Also, what factors might cause quality shifts? Also, how might income, taste and dietary changes affect which grain types and varieties will be preferred in the future?

Image: baked goods are becoming more popular in Indonesia as demographics change

We know that a country’s demographics are a prime indicator of whether it will demand more or less. In general, a growing population indicates more mouths to feed. Growing incomes usually lead to an increasing appetite for high quality food, and potentially a shift in the types of food consumed. Sources of this type of data are many and are relatively easy to access and interpret.

However, second-tier effects, such as growth in the middle class, trending health issues, food safety or environmental concerns are more complex indicators; yet their effects on demand can be powerful. A prime example is the Renewable Energy Directive in the EU, which required suppliers of canola to prove their sustainability. Had Australia’s canola industry, with help from AEGIC, not responded in an effective and timely manner to show its sustainability credentials, then Australia would not have had access to the highly-lucrative EU biodiesel market. At the time this market was worth over $1 billion. These second-tier concerns or indicators may still be emerging, and their importance is not easily shown in predictive tools like regression analysis that rely on historical datasets.

In such cases where new factors are emerging, a better approach to prediction may be the generation of ‘fuzzy’ data, supplied by accessing many well-informed industry insiders who share their interpretation of what is happening in a market. From their combined views are distilled emerging yet still uncertain trends, using what is known as fuzzy logic which provides ways for handling vague and imprecise information (D’Urso 2017). This method for generating data can be employed to create indices such as the global food security index (GFSI 2021) or the global competitiveness index (World Economic Forum 2020).

This approach has some merit and has been shown to generate more accurate forecasts than a single expert undertaking their own analysis (e.g. Graham 1996). The Delphi method is a similar approach, also grounded in the belief that a group of well-informed participants is more useful than a single view (Hsu 2007) and this approach can also be expanded into a ‘fuzzy Delphi’ method (Habibia, 2015).

While an expert group is unlikely to hold a consensus view, it is likely that there will be a shared view on the importance of common drivers in markets and the relative importance of those drivers. Without this information, market forecasts rely only on observed data, and thus often have a reliable forecast range of only a few years. Such time frames are insufficient for helpfully informing value chain participants who need to make long-lived investments, such as developing a new variety and who face long lags between investments and returns. Businesses who need to make such investments need accurate long-term projections. Methods like fuzzy logic and fuzzy Delphi are methods to aid such decision-making.

AEGIC’s Economics and Market Insights Team is looking to employ such methods in a future project that will identify market opportunities and gaps that could benefit the Australian grains industry.

Sources

Global Food Security Index (2021) The Economist Group, https://foodsecurityindex.eiu.com/

Arash Habibia, Farzad Firouzi Jahantigh, Azam Sarafrazi (2015) ‘Fuzzy Delphi Technique for Forecasting and Screening Items’, Asian Journal of Research in Business Economics and Management Vol. 5, No. 2, February 2015, pp. 130-143

Hsu, Chia-Chien and Sandford, Brian A. (2007) ‘The Delphi Technique: Making Sense of Consensus’, Practical Assessment, Research, and Evaluation: Vol. 12 , Article 10.

John R. Graham (1996) Is a Group of Economists Better Than One? Than None? The Journal of Business Vol. 69, No. 2 (Apr., 1996), pp. 193-232 (40 pages)

World Economic Forum (2020) Global Competitiveness Index, Reports | World Economic Forum (weforum.org)