Individuelle Daten zu Kurzarbeitenden: Datenvalidierung und erste Befunde

Abstract

"Short-time work has proven to be a good instrument for securing employment in crises, most recently during the coronavirus pandemic, when its use reached an unprecedented peak in April 2020 with up to six million people being in short-time work simultaneously. However, very little is known about which people receive how much short-time working allowance. Short-time work is initiated by establishments, but the establishments’ notification is accompanied by an accounting sheet listing all employees (subject to social security contributions) who are in short-time work. During the pandemic, information on the individual receipts of short-time working allowances for the period from March 2020 to December 2021 was automatically extracted for the first time, using an automated optical character recognition procedure. We show that many characters could not be recognized by the program and are therefore missing. In addition, there are a range of implausible values. After cleaning the data, we validate the newly generated data by comparing it with establishment-level short-time work data and existing administrative employee data sets. This allows us to generate a foundation of high-quality short-time work data. We then use marginal distributions for the number of people in short-time work (taken from the statistics department of the Federal Employment Agency) to carry out a weighting procedure (namely, iterative proportional fitting). Specifically, for each month, the marginal distributions by gender, establishment size, economic sector and region as well as the total work loss are taken into account. Using the weighting allows to make representative statements, also with regard to previously unknown dimensions of short-time work, e.g. by level of education during the pandemic. We show that the weighting procedure works well in this case (e.g. it produces weight distributions without outliers) and has a high degree of flexibility. In addition, the weighting shows, for example, that employees from large establishments with 250 or more employees are underrepresented in the data on the individual utilization of short-time work benefits. This selection appears to be due to the fact that the text recognition procedure aborted the extraction of data from long notification lists after an unspecified number of pages. Using the new- data and the weighting procedure, we calculate a series of new descriptive results on the use of short-time work during the pandemic. We focus on two measures: The ratio of short-time workers to all employees within a group and the average wage loss of short-time workers within a group. The average wage loss describes the share of wages which is lost due to short-time work and which is partially reimbursed by the Federal Employment Agency. Overall, the results show that low-skilled and low-income workers were significantly more affected by short-time work. Firstly, the short-time work rate was higher in these groups and, secondly, they had a higher average wage loss when in short-time work. At the short-time work peak in spring 2020, around 19 percent of all employees subject to social security contributions were in short-time work, with an average wage loss of 50 percent. The rate of employees in short-time work was 22 percent for those without a vocational degree, but only 14 percent for those with a university degree. The differences based on gross monthly pay were even greater for full-time employees: A third of all full-time employees with gross wages of less than 2,000 Euro were in short-time work in April 2020, but only 14 percent of employees earning wages higher than 4,000 Euro. At the same time, employees with gross monthly wages of less than 2000 Euro had an average wage loss in the event of short-time work that was – at more then 50 percent on average – 20 to 25 percentage points higher than for those earning wages higher than 4000 Euro. Another result of the analysis is that the sharp increase in short-time work at the turn of the year 2020/2021 is mainly driven by employees with a low level of education and relatively low wages. Looking at occupational patterns, we show that the short-time work rate in spring 2020 was highest in production occupations. Thereafter, it was higher in service occupations, especially in personal services, where the resurgence of short-time work during the second lockdown was particularly pronounced. In the case of short-time work, the average loss of wages is however higher in service occupations across the entire period. In conclusion, we note that - despite the incompleteness of the data - the new individual data on short-time work offer further potential, e.g. with respect to longitudinal analyses." (Author's abstract, IAB-Doku) ((en))

Cite article

Kagerl, C. & Kruppe, T. (2024): Individuelle Daten zu Kurzarbeitenden: Datenvalidierung und erste Befunde. (IAB-Forschungsbericht 05/2024), Nürnberg, 28 p. DOI:10.48720/IAB.FB.2405