Fig 1.
Note: This figure illustrates the data framework for tracking effects of economic shocks on businesses by combining corporate website, business survey and credit rating data.
Fig 2.
Companies with COVID-19 references on their corporate websites after announcement of first economic shutdown.
Note: Figure shows the number of firms which reported about COVID-19 on their corporate websites over time, shortly after the announcement of the first nationwide shutdown at March 16, 2020 (left vertical axis). The repeated design of the web queries allowed to monitor the near real-time impact of the pandemic on the corporate sector. Red line (right vertical axis) depicts the growth rate of companies reporting about COVID-19 on their websites. Growth rate is calculated on a rolling basis with window size 3. Fluctuations towards the last few web queries both reflect an improved scraping process that was implemented in early May 2020 and companies that have removed COVID-19 references from their websites.
Table 1.
Distribution of context classes in the training data.
Table 2.
Performance of context classification on test set.
Fig 3.
COVID-19 firm communication on corporate websites website-generated impact values at sector level.
Note: Visualizations based on classified COVID-19 web references. If a firm reported at least one COVID-19 reference that has been classified in the respective context class in any of the web queries, the firm gets assigned a 1. Else the firm gets assigned a 0 for the respective class (binarized version of the web indicators). Red lines represent sector-specific impact values. Grey shaded areas represent unweighted average impact values across all sectors. Exact numerical values can be found in S6 Table.
Table 3.
Descriptive statistics: Corporate website data.
Table 4.
Descriptive statistics: Survey data.
Fig 4.
COVID-19 firm exposure at sector level based on survey results.
Note: Visualizations based on survey questions A—F which were asked on a 0–4 Lickert scale with 0 indicating no negative effects, 4 signaling strong negative effects. Red lines represent sector-specific impact values. Grey shaded areas represent unweighted average impact values across all sectors. All values are averaged at the firm-level across the three survey waves.
Fig 5.
COVID-19 effects on corporate solvency at sector level.
Note: Figure shows distribution of credit rating updates both during COVID-19 (yellow to red palette) and before COVID-19 (dashed green line). Densities are based on a Gaussian smoothing kernel with a bandwidth of 2.
Table 5.
Descriptive statistics: Credit rating data.
Table 6.
Distribution of extreme rating downgrades.
Fig 6.
Comparison webdata and survey data effect estimates.
Note: Figure shows average marginal effect estimates and corresponding 95%-confidence intervals of model 3 where the dependent variable (negative impact) is generated from webdata (green) and survey data (red). Dependent variable from webdata reflects whether the firm has reported a ‘problem’ reference on its corporate website in any of the web queries. Dependent variable from survey data refers to the question whether the firm has suffered negative impacts due to the pandemic in any of the three survey waves. Shaded estimates signal statistically insignificant effects at the 5% level. Incumbent firms (10 years and older) serve as baseline age group, micro-enterprises (number of employees ≤ 10) as baseline size group, accommodation and catering serves as baseline sector among the sector dummies. Marginal effects need to be interpreted relative to the baseline group(s).
Table 7.
Regression results: COVID-19 references on corporate websites as early indicators for changes in firm credit ratings.