4 What-if Analyses
Times of Uncertainty
By John Pasinski, March 30, 2020
When the environment we operate in changes, one question analytics leaders face is whether or not their previous work is still relevant going forward.
Are the models you built yesterday still useful today?
Undoubtedly, the implicit or explicit assumptions built into analyses do not hold up over time. The reality is that the world is always changing, sometimes very rapidly. Analytics leaders face the challenge of ensuring that their modeling work reflects what is happening today and likely to happen going forward. In order to achieve that end, analytics leaders need to have a strong understanding of their assumptions and how they impact their models.
An effective way to incorporate changing dynamics in analytical efforts is to have models that incorporate cause and effect. Causal modeling allows analytics teams to explore the processes that generate data, as opposed to just patterns in the data itself. When the world evolves, examining historical patterns and correlations in data is necessary, but not sufficient to gain a better understanding of what is likely to happen going forward.
Simulation is a popular causal modeling approach in today’s current environment because it helps to reveal why certain outcomes will happen through what-if analysis. The Washington Post used simulation to explain why outbreaks like coronavirus will spread and how to flatten the curve. FiveThirtyEight, a political website, used simulation to forecast who will win the Democratic primary, while MIT is using simulation to find solutions for climate change.
During this time of uncertainty, econometric models and AutoML are not equipped to analyze and forecast dramatic shifts in human and market behavior. There is a need to create a unified view of a market in order to understand the cause and effects of strategic decisions.
Market simulation uses complexity science, behavioral economics, agent-based modeling, network science and artificial intelligence to make this possible. It is among the only bottom-up methodologies that enables the analysis of market conditions in which the micro-behavior of individual people is the primary force of macro-outcomes.
Here are four areas where market simulation is being used to understand and forecast human behavior:
1) What-if Demand Changes
There are always events that may alter demand in different industries. Changes in technology and consumer preferences can shift markets – think about how CDs and DVDs have become increasingly obsolete. Weather events can change demand – fewer people may go out to open new bank accounts if there is a large winter storm, but on the other hand sales in snow shovels will go up. Economic ups and downs can shift demand as well.
2) What-if Distribution Changes
Distribution can shift for a variety of reasons. Retailers may stop carrying specific products or start stocking new items. Ecommerce has been opening up new channels for consumers to gain access to different products. Trade wars can disrupt the complex supply chains for products like electronics, appliances, or automobiles and lead to shortages and delays. Competitors may change their distribution strategy as well. Just as with demand, there are a number of factors that can lead to shifts on the supply side.
3) What-if Consumer Preferences Change
Consumer preferences also shift in response to a variety of factors. Trends in pop culture, fashion, and style are always evolving. Many consumers are becoming more eco-conscious when choosing the products they buy — being more cognizant of factors like recyclable or compostable packaging or whether their products are locally sourced. As the economic environment changes, individuals may become more or less price sensitive. Consumer preference changes influence the dynamics of markets and how consumers make choices.
4) What-if Consumer Perceptions Change
Significant events often lead consumers to pay more attention and respond more acutely to the actions companies take. When an Icelandic volcano eruption disrupted air travel across the Atlantic years ago, some travel brands took a hit for bungling the response while others navigated the issue more effectively. There are a number of types of events that can shift consumer perceptions. When consumers get sick due to issues with food safety at restaurants, those restaurants’ reputations and brand health are likely to be changed. Shifts in consumer perceptions can impact consumer behavior in new ways that are not always easy to anticipate.