As we have already seen in the last six months, the effects of COVID-19 on businesses have differed dramatically. While recreational vehicles, pet care, and home improvement have skyrocketed, live entertainment, restaurants, and bars struggled immensely during lockdowns.
Given these remarkable differences across industries, a narrower approach to forecasting is becoming more and more critical. In a recent Q&A, Prevedere CEO Rich Wagner and Dr. Barry Keating, professor at the University of Notre Dame, stressed the importance of micro forecasting with real-time data to capture changes across industries and regions.
Given the unique effects of the current recession and COVID-19 on businesses, we can no longer use the same time series models for economic forecasting as we have in the past. Data-driven forecasting needs to be grounded in current patterns rather than the older models from before the pandemic. As Dr. Keating explains, “there are still patterns, but you can’t look at the patterns as you did five years ago. You have to look at the patterns five minutes ago, and five days ago, and five weeks ago.”
Luckily, there is a world of individualized, real-time data becoming increasingly available. As a result of COVID-19, more types of data are becoming more easily accessible. One example of this is mobility data, such as mobile phone and social media data. This type of real-time data is crucial to understanding and predicting trends in the next few months. While the pandemic and current economic situation are undoubtedly unprecedented, real-time data allows forecasting to be more accurate, more up-to-date, and more responsive to the rapidly changing economic environment.
Real-time data is one aspect of successful economic forecasting, but it is increasingly important to consider regional differences. Typically, we would expect that states would have similar trends of downturns and recovery during a recession. However, from the beginning of the pandemic, we have seen a wide variety of government responses and shutdown levels, even sometimes by neighboring states. Rich Wagner suggests adopting “a more micro focus where we’re looking at more granular data, more near-term data, but also external data around the state’s behavior.”
We can no longer look at a region like the Midwest during forecasting because what’s happening in Michigan or Ohio may differ from in Indiana or Illinois. Indicators from one state may not be useful for understanding what’s happening in another. Notably, the first states to go into lockdown in response to COVID-19 and the states with the strictest mandates are facing a slower economic recovery than others.
COVID-19 has forced us to evolve and adapt our forecasting models. A more micro approach that focuses on real-time, granular data provides more accurate and specific forecasting outcomes. Specifically, this type of data-driven analytics can inform successful scenario planning and help businesses to not only react and adapt to future changes but learn to thrive in the face of uncertainty and become more resilient.