Short-term, agile planning proved to be best approach during pandemic supply chain disruption
Most manufacturers prepare for supply or demand disruptions by holding extra inventory. The COVID-19 pandemic disrupted both supply and demand, with some firms seeing sharp declines in demand while others saw unforeseen spikes.
While previous research suggests steps a firm could take in such situations, their success had not been evaluated in a disruption as severe as the COVID-19 pandemic, said Jason RW Merrick, Ph.D., a professor of supply chain management and analytics at the Virginia Commonwealth University School of Business.
Merrick is a member of the Advanced Supply Chain Collaborative, a University of Tennessee organization of companies and academics that researches hard problems facing companies’ supply chains. In September 2019, the group began researching ways to build agility into existing supply chains with its research partner, a packaged food company. In March 2020, as COVID-19 hit, the company faced a large increase in demand and the threat of running out of stock if it did not react quickly.
“When Preemptive Risk Mitigation is Insufficient: The Effectiveness of Continuity and Resilience Techniques During COVID-19,” published in the journal Production and Operations Management, shows that decision-making changes — specifically changes to the cadence of planning and the forecast time horizon — had more impact on business continuity than those focused on resource reconfiguration. The research provides insights that are generalizable to many firms as uncertainty persists in global supply chains.
Merrick discussed his research with VCU News.
Why is this research important? What can we learn from it?
Most products are made by global supply chains that are based on long-range forecasts and long-term planning. The physical machinery used to manufacture these products is often extremely expensive, so it must be used efficiently to turn a profit. This type of long-range planning is built into the processes and the software that run their supply chains. But efficiency is often at odds with agility. The disruption of COVID-19 on our global supply chains required companies to turn on a dime, and their efficient supply chains were not ready.
There is a lot of research that suggests ways to mitigate disruptions, like holding extra inventory to get you through the crisis, shortening your forecast horizon, making shorter-term plans (allow the supply chain to change its mind), adding extra physical capacity, or reducing the variety of products you are making. However, there is no research that says which measures would be effective against a disruption as widespread and significant as COVID-19.
Our research partner firm saw a large increase in demand in March 2020, and their dam of extra inventory would be quickly wiped out if they didn’t act quickly.
Can you explain how the pandemic disrupted the supply chain in both directions (up and downstream)?
Some companies saw demand for their products fall off a cliff, while others saw large increases in demand. Our research partner saw demand for their restaurant products disappear, while their demand for regular products increased by almost a third. In normal times, they might see a one-third growth occur steadily over multiple years. In March 2020, it happened in a couple of weeks as those of us who could went into quarantine. Supply for certain types of ingredients and raw materials dried up, while others remained steady. They could not build new production lines to keep up with this demand and they did not know how long the increase would last.
They quickly put in place measures to keep their workers safe at their plants, but how could they keep up with such high demand? Their answer was to try everything mentioned in the research literature to deal with disruptions. This gave us a perfect real-world test to figure out which combinations of measures worked and which didn’t.
Describe the experiments you conducted.
We built a simulation model of their supply chain for the year leading up to the COVID-19 disruption and the first three months — March, April and May 2020. We made sure that the pre-disruption simulation accurately represented their normal operations and its output matched their inventories and service levels for that year. We also made sure that the disruption simulation accurately represented the new measures they put in place — including when they started each — and its output matched their inventories and service levels for that period. This made sure that our simulation was a valid representation of their operations both pre- and post-disruption.
Next, we ran experiments to see which combinations of measures allowed them to recover most quickly. Our model accurately represented the effect of each measure and how the measures interacted, so we had a test-bed for determining the best actions. The four individual measures were:
- Reduce the horizon of their forecasts to one week instead of their previous one-month forecasts for finished goods demand and four-month forecasts for planning the purchase of ingredients.
- Plan production for a week at a time, instead of their previous monthlong plans.
- Run the production lines six days a week, instead of their previous five days a week.
- Produce only their top 15 products instead of their normal 158 products.
What conclusions did you come to?
We found that planning production one week out and only forecast for that one week ahead was the most effective combination. They could no longer use long-term plans to be efficient. Their customers’ world was changing, so they had to evolve with it. This kind of change was extremely hard. They needed to shelve their existing processes and plan using spreadsheets instead of their existing planning systems. They needed to create new processes and work long hours.
The simulation also showed that they did not need to run every Saturday if they were planning properly, and they could have made all of their products instead of cutting down their selection. However, in discussions with the managers we did call into question this last conclusion. The computer model showed that they could have kept up with the demand for all their products, but the planners’ workload would have increased even further and they were already stretched. They may have made mistakes or burned out.
What would you like to add?
In our research, the two measures that worked together (shortening the forecast horizon and increasing the cadence of planning) are termed cognitive measures because they affect the decision-making at the firm. The two that were shown to be less important (adding production capacity by working every Saturday and decreasing the diversity of their products) are termed physical measures because they change the physical supply chain. Many firms jumped straight to physical measures, but the cognitive measures were the most effective.
What about the next major disruption? As demand returned to more normal levels, the firm could return to the more efficient long-range planning. But they should be ready to throw the switch back to short-term, agile planning at the next disruption. As we learned from COVID-19, this isn’t easy and supply chains need to be designed to be efficient when efficiency is possible and agile when a disruption occurs.