Redefining the Science of Better

How much raw material should you purchase given the uncertainty in demand and procurement cost? SMU Associate Professor Onur Boyabatli can distil your business environment into an analytical model that can answer these questions.

Back to Research@SMU Issue 35

Photo Credit: Cyril Ng 

By Chin Wei Lien & Chua Kim Beng

If you are a farmer contemplating how much land to allocate between two crops — say, soybean and corn — what factors will you consider?

An important point to think about will be the revenue each type of grain can earn. You will plant more soybean if it earns you more money than corn. However, by the time the two crops are ready for harvest, their revenues may have swapped positions – a phenomenon that is commonly observed in practice owing to the commodity nature of these crops.

And what if you grew soybean in your land the last season? There are well-known benefits of growing a different crop in the same land, such as crop rotations, which improve the soil structure and break the pests’ reproductive cycles. This will make it less appealing to allocate more land for growing soybean this season.

“If we continue discussing, we can identify a lot more factors,” advises Onur Boyabatli, Associate Professor of Operations Management in the Lee Kong Chian School of Business at the Singapore Management University (SMU). “For example, does the farmer have sufficient resources such as fertilisers and pesticides for cultivation, and labour and machines for harvesting?”

Professor Boyabatli works on analytical models that take into account these variables to answer the farmer’s questions about land allocation. But models like these do not apply to farmers alone. “My main research is on Operational Decision-Making in Commoditizied Industries, with the special interest in the agri-business,” he elaborates. “What I do is to look at different supply chain agents in agricultural industries, such as a farmer, a processor, a refinery. Then I examine the operational problems among these agents, for instance, capacity investment, procurement or product pricing. These are all operational areas that require decisions.”

But, being in Singapore, why did Professor Boyabatli choose to study the agri-business, when this country is highly urbanised? “Singapore is a very big agriculture trading hub,” he discloses. “There is not much agriculture going on in Singapore, but there is a lot of agricultural trading, plus a lot of agriculture in Southeast Asia.” Agricultural activities, such as the palm industry within and around Singapore, have an influence on Professor Boyabatli’s research. The earlier case on crop allocation was an illustration from his paper “Crop Planning in Sustainable Agriculture: Dynamic Farmland Allocation Corn or Soybean? Dynamic Farmland Allocation in the Presence of Crop Rotation Benefits”. Professor Boyabatli has also applied his analytical models to optimise the procurement process for processors (i.e., meatpackers) in the beef industry, as in the case of Integrating Long-Term and Short-Term Contracting in Beef Supply Chains.

Building the model

Professor Boyabatli’s main task starts with understanding the business processes, deciding on the important variables, and then developing a quantitative model that considers all these facets to answer particular questions. “By using applied mathematics and probabilistic analysis, my goal is to come up with some managerial insights,” he explains.

With his analytical model, he can advise a farmer on what actions to take should a particular event occur. “Let’s say the price of corn increases. What adjustments should the farmer make?” he questions. “How much are the crop rotation benefits? Is it really significant and should the farmer consider them?” Using the model, I can say, “Look, in your case, rotating crops increases your profit by 20%. That’s really important.”

So what factors need to be considered when setting up an analytical model?

“There is always a trade-off,” Professor Boyabatli says. “Ideally, we want to put every factor in the model to make it more realistic, but this increases its complexity, making it harder to solve.” And even if the complex model is solved, he adds that it will still be hard to extract managerial insights, “because there are so many moving parts.”

Hence, it is important to establish a hierarchy of factors to consider in the model. “Ask the farmers to prioritise,” Professor Boyabatli suggests. He also shares that he is not the only person working in this area. “There is a big community of academics studying economics of agriculture. I look at what other people have developed and build on what they have done. That also helps me determine which factors are important and which ones are not so important.”

Calibrating with real-world data

Once Professor Boyabatli comes up with an analytical model, he calibrates it using real-world data to represent a typical decision-maker in practice. Quoting the example on crop allocation, he says, “By keying in crop prices, crop yields, rotation benefits for a typical farmer in, say, Iowa, my model gives a recipe. That recipe, however, may not be suitable for a farmer in Minnesota because the environment is different.” To make sure that his findings are robust, Professor Boyabatli solves his models without making any assumptions on parameters. Calibrating these models using real-world data enables him to generate additional managerial insights.

Model calibration is not as easy as it sounds. “It is made difficult because you need to find the data,” says Professor Boyabatli. He considers himself fortunate that such data for agricultural commodities are easily available in Singapore. However, such data are regarded as  trade secrets in numerous other sectors, thus making calibration difficult, if not impossible.

Another difficulty is the extra analyses required for generating insights after calibration. “In many fields, this is called numerical analyses,” he discloses. “So when you build the analytical model, you need to conduct simulations to explain them.” These analyses require complex tools to fit the model components to data and to simulate the calibrated model.   

The farmer does not need to understand the complex tools involved in the analysis. “All they need to know is what they are supposed to do; I’ll be the one explaining to them why,” he says. “If your corn prices are expected to increase more significantly than your soybean prices, then you should plant more corn. But how much more? Well, I’m going to tell you exactly how, based on the model.”

Academic background

Professor Boyabatli, who hails from Turkey, started his academic career studying Industrial Engineering. “I did my undergraduate and Masters in Industrial Engineering in Bilkent University,” he reveals. “Bilkent is one of the best universities in Turkey, with a very strong engineering school. The education is very academic, and a lot of graduates directly go to the top PhD programmes in the US and Europe.” He then enrolled in INSEAD, in France, one of the top business schools in the world, to take a PhD degree in Technology & Operations Management. His research interests are in the area of integrated risk management in supply chains. Besides agri-business, he also studies technology and capacity management under financing frictions in capital-intensive industries (such as automotive); and operational and financial hedging in global supply chains.

Professor Boyabatli is also the co-editor of “Handbook of Integrated Risk Management in Global Supply Chains”.

Back to Research@SMU Issue 35