ISTARI Empowers German Government with Real-Time Economic Insights During COVID-19.
In a nutshell
We developed an innovative time-series analysis framework that integrated real-time web data, business surveys, and credit ratings. By utilizing advanced Natural Language Processing (NLP) techniques on corporate websites, we enabled the German Government to swiftly identify and understand sector-specific impacts of the COVID-19 pandemic. This cutting-edge approach provided timely and actionable insights, allowing for more targeted and effective economic interventions, significantly enhancing the decision-making during the COVID-19 pandemic.
What was the challenge?
How can policymakers make informed decisions when traditional data sources lag behind rapidly evolving economic crises? At the onset of the COVID-19 pandemic, the German Federal Ministry of Economic Affairs and Energy faced this dilemma. Official statistics and surveys were too slow to capture the swift and varied impacts of the pandemic on different industries, leaving decision-makers with delayed and aggregated information. There was a critical need for real-time, granular insights to effectively support decision-making.
How did we help?
In the first step, we analyzed over a million German corporate websites to detect mentions of COVID-19. Utilizing our advanced NLP models, we categorized these references into contexts such as problems faced, adaptations made, or general information sharing, providing an early snapshot of how different sectors were responding to the pandemic.
In addition, targeted surveys complemented the real-time web data with nuanced, firm-level perspectives. Moreover, we integrated credit rating data to assess the long-term financial health of companies. By linking early web indicators with subsequent credit rating changes, we demonstrated how initial distress signals could predict future financial solvency, offering a comprehensive view of the economic landscape. Our near real-time assessments of firms' affectedness was also confirmed by their credit ratings which became available one year later.
Read the full paper here: https://doi.org/10.1371/journal.pone.0263898
How did we add value?
By delivering real-time insights on a regular basis, we enabled policymakers to act swiftly and accurately. Our time-series analysis highlighted which sectors were most in need, allowing for targeted economic support that maximized impact and optimized resource allocation. This not only enhanced the effectiveness of policy measures but also minimized unnecessary fiscal burdens, ensuring that assistance was both timely and efficient.