The high frequency firm survey 'Bundesbank Online Panel-Firms'
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Boddin, Dominik; Köhler, Mona Article The high frequency firm survey 'Bundesbank Online PanelFirms' Journal of Economics and Statistics Provided in Cooperation with: De Gruyter Brill Suggested Citation: Boddin, Dominik; Köhler, Mona (2024) : The high frequency firm survey 'Bundesbank Online Panel-Firms', Journal of Economics and Statistics, ISSN 2366-049X, De Gruyter Oldenbourg, Berlin, Vol. 244, Iss. 3, pp. 267-275, https://doi.org/10.1515/jbnst-2023-0009 This Version is available at: https://hdl.handle.net/10419/333283 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/
Data Observer Dominik Boddin* and Mona Köhler The High Frequency Firm Survey “Bundesbank Online Panel –Firms” https://doi.org/10.1515/jbnst-2023-0009 Received January 26, 2023; accepted January 27, 2023 Abstract: The Bundesbank Online Panel –Firms (“BOP-F”) is a dataset with responses from a high frequency firm-level survey of the same name. The Bundesbank has conducted the survey since June 2020, and since July 2021 the survey has been carried out at a monthly frequency. Every month, around 3000 firms from all economic sectors, regions and size classes are surveyed. The survey consists of recurring core questions about the economic situation of firms and their expectations and special questions that usually differ from quarter to quarter. The latter often relate to current topics, for instance, climate change, digitalization, Covid-19. The data can be accessed for research and especially the possibility to combine it with other administrative Bundesbank data makes it particularly valuable for research. The objective of this paper is to describe the methodology of the data collection, the content data as well as the data’s research potential. Keywords: establishment survey, survey methodology, corporate finance, COVID-19 pandemic, labor market, firm decisions JEL Classification: C81, C83, G30, M21 Any opinions expressed in this paper represent the author’s personal opinions and do not necessarily reflect the views of the Deutsche Bundesbank or the Eurosystem. The content of this paper corresponds in part to the content from the data report of the data, cf., Boddin et al. (2022). We thank the entire BOP-F survey team (Stefan Bender, Satyajit Dutt, Julia-Katharina Ginz, Heba Ismaeil, Sabine Lösch, Tobias Schmidt, Pawel Smietanka, Milena Tzvetkova) at the Deutsche Bundesbank. *Corresponding author: Dominik Boddin, Deutsche Bundesbank, Wilhelm-Epstein-Strasse 14, 60431 Frankfurt, Germany, E-mail: [email protected] Mona Köhler, Deutsche Bundesbank, Wilhelm-Epstein-Strasse 14, 60431 Frankfurt, Germany, E-mail: [email protected] Journal of Economics and Statistics 2024; 244(3): 267–275 Open Access. © 2023 the author(s), published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License.
1 Introduction The abrupt and sharp economic collapse caused by the Covid-19 pandemic in spring 2020, triggered a great need for detailed up-to-date information on the economic situation of firms in Germany. The Research Centre and the Research Data and Service Centre of the Deutsche Bundesbank therefore initiated its own online firm survey, the Bundesbank Online Panel –Firms (“BOP-F”), in particular to obtain information on the pandemics effects on the general economy, on the financial situation of firms and on future expectations in the corporate sector. Since July 2021 until today, the survey is carried out monthly with around 3000 participating firms and has established itself as an important tool providing valuable information for analysts and researchers. 1 The survey data for each month represent a cross-sectional data set. The cross-sectional data from three consecutive months are combined into a quarterly unbalanced panel. Due to the underlying rotating survey procedure with survey of the panel firms every quarter as well as unchanged questionnaires within a quarter, BOP-F contains quarterly survey data with an unbalanced panel structure. Quarterly weights are provided accordingly. Every quarter, around 9000 firms from all economic sectors, regions and size classes are surveyed. The survey consists of recurring core questions about the economic situation of firms and their expectations (e.g. price development or inflation rate expectations) and special questions that usually differ from quarter to quarter. The latter often relate to current topics. While initially Covid-19 was the main focus, meanwhile a large number of topic areas with current relevance were covered. This includes but is not limited to climate change, digitalization, minimum wages, supply chain issues, energy prices etc. Researchers can access the anonymized data at the RDSC’s guest researcher workstations in a secure environment at one of the RDSC’s locations. One big advantage of BOP-F data is that it can be combined with other administrative Bundesbank data, which makes it particularly valuable for research. The remainder of this paper is organized as follows. Section 2 introduces the methodology of the data collection, i.e. in particular sampling and weighting, and Section 3 presents the content of the data. Section 4 describes the data access and the opportunities for research, especially through matching with administrative data. Section 5 offers some concluding remarks. 1The survey began with a one-year “pilot phase”during which five online surveys were conducted approximately every two months. In addition, before the start of the pilot phase, a pre-test survey with a sample size of 1000 companies was carried out in order to test the questions and the survey design and to obtain initial indications of the response rate and willingness to participate in the panel. 268 D. Boddin and M. Köhler
2 Data Methodology This section contains a description of the methodology including the survey design, the data collection process and the weighting procedure. For further details please check the data report (cf., Boddin et al. 2022). BOP-F is a dataset based on a representative online survey of firms conducted by the Research Centre and the Research Data and Service Centre (RDSC) of the Deutsche Bundesbank in cooperation with an external survey firm, forsa. BOP-F contains quarterly survey data collected monthly over three-month cycles. 2 The survey structure is based on a rotating panel principle, as shown in Figure 1. This structure allows us to compile representative data for the population of firms based in Germany on a monthly basis and to reduce survey fatigue for firms. While the BOP-F survey is conducted every month, the questionnaires only change after the end of a quarter. The sample –consisting of both panel and newcomer firms, 3 is drawn for an entire quarter and is then divided evenly over the three months. However, once a firm has participated in the survey in the first month of the quarter, it is always invited to participate in the first month of every quarter going forward. In Figure 1, for instance, firms in group 1 will always receive the questionnaire in the first month of every quarter, i.e. January, April, July and October. With this rotating panel design, we keep the burden on firms as low as possible by contacting a given firm only four times a year. However, by conducting the survey monthly, high-frequency monthly data is generated. Since not all companies are willing to take part in the survey again and the number of participants per wave should be kept comparatively stable over time, the survey sample must be refreshed regularly. The sample for the BOP-F survey is drawn from the universe of firms located in Germany with a taxable turnover of more than €22,000 4 or at least one employee paying social security contributions. 5 The sample for every quarter is drawn proportionally to the distribution of the population of firms along three dimensions: (i) a firm size indicator measured in 2As previously mentioned monthly data is available from July 2021 onwards. Prior to that the survey was conducted at lower frequency. 3We differentiate between firms that participate in the survey for the first time (“newcomer firms”) and those who have participated before (“panel firms”). 4In the first waves the boundary for taxable turnover was slightly lower at €17,500. 5This data base is frequently updated, but the cut-offreporting date on which the data is based on is around two years before the actual survey. Firms which are not classified as natural persons or for which information such as address, sales, or employment is missing are excluded from the sample. Around 12% of the original database could not be used due to missing information. The population contains only enterprises that are registered in Germany. The total available population contains around one million firms. The “Bundesbank Online Panel –Firms”269
terms of employment and turnover (3 classes), (ii) region (4 classes), and (iii) economic sector (6 classes). Thus, the sample is drawn from 72 strata (3 ×4×6). Afterwards, the quarterly sample is distributed equally over three months. 6 Due to the disproportional sampling design of the BOP-F survey, if unweighted, data analysis would lead to non-representative results. In order to make the results representative of the universe of German firms weights are created along the three dimensions mentioned above. 7 The field work of each wave usually starts on the first business day of a month and ends around 22 days later. 8 After about half the time of the field phase, firms are reminded to participate in the survey. Only firms that have not responded by then receive reminders. One day before the field phase ends firms receive a final Figure 1: Rotating panel design. 6The sectoral classification corresponds to the IAB’s Establishment Panel classification and is similar to NACE Rev. 2. 7In the first step, design weights were created by using the inverse of the inclusion probabilities. Then, sampling weights for each stratum are iteratively adjusted by using an iterative proportional fitting procedure. The aim of this procedure is to adjust the sum of weights for each marginal dimension of the target variables to the distribution of the population until a predefined convergence criterion is achieved. The target structure is based on three dimensions: region economic sector and firm size class. 8The first five rounds of online pilot surveys were conducted approximately two to three months apart and a field phase lasted for about 20–40 days. 270 D. Boddin and M. Köhler
reminder. This second reminder is sent via email only and thus only to panel firms who provided their e-mail-address. To ensure that highly ranked executives fill out the questionnaire, the phrase “for the attention of the management”(in German: “zu Händen der Geschäftsführung”) is printed on every envelope with survey invitations. Before each survey, respondents are informed that the survey would take around 15–20 min to complete. The actual median interview duration for uninterrupted interviews lies within this time interval. Around 60% of the newcomer firms state that they are willing to take part in the future surveys. The actual future participation rate, however, is lower. Table 1 summarizes the key figures discussed in this section for each wave. The first column states the wave and the second column the duration of the field phase. Column three contains the number of firms that have been contacted to participate in the survey, whereas column four contains the number of successful interviews that corresponds to the number of observations in our data. As mentioned previously, the monthly design with a smaller number of observations was introduced in July 2021. Column five contains the response rate in percent, which is the share of participated over contacted firms. Column six contains the share of panel firms, i.e. firms that already had participated in the survey prior to that wave. The remaining firms are newcomer firms. Column seven states the median duration that firms required to finish the survey. As already mentioned, during the pilot phase lasting until wave five, the sample size was significantly larger and the field phase longer. Since the introduction of the monthly frequency in wave 6, both the gross and the net sample size have been stable. Currently, (as of December 2022) 43,189 different firms participated in the survey which lead to 104,270 finished interviews and thus observations in the data. Also the response rate is rather stable at around 15% to 20%. Not surprisingly, the share of panel firms has steadily increased over time. Currently, more than 80% of firms that take part in the survey have done so before. 3 Content of the Data The questionnaires in the BOP-F study comprise a set of core and special questions. Core questions are included in every survey wave, change little from one quarter to another and ask about the economic situation of firms and their expectations. In general, the core questions cover the three main areas (I) macroeconomic questions, (II) microeconomic questions related to real decisions, and (III) microeconomic questions related to financial decisions. Some of the core questions are probabilistic and allow measuring firms’uncertainty regarding their external environment (e.g. the key interest rate) and their own future key figures and outcomes (e.g. sales). The “Bundesbank Online Panel –Firms”271
The core questions are supplemented by special modules that differ from quarter to quarter. They are designed for more in-depth investigation of certain topics and are often about important current issues. The main focus of the special questions prior to July 2021 (i.e., during the first five survey waves) was the measurement of the impact of the Covid-19 pandemic on firms. These include, for Table :Key figures by wave. Wave Field phase duration Gross sample Net sample Response rate (in %) Panel share (in %) Duration in minutes June –July , , . –. August –September , . . . October –November , , . . . January –March , , .. . May –May , . . . July –July , . . . August –August , . . . September –September , . . . Ocotber –October , . . . November –November , . . . November –December , . . . January –January , . . . February –February , . . . March –March , . . . March –April , . . . May –May , . . . June –June , . . . July –July , . . . July –August , . . . August –September , . . . October –October , . . . October –November , . . . November –December , . . . 272 D. Boddin and M. Köhler
example, the impact of the temporary VAT reduction, the analysis of potential liquidity bottlenecks, and the pandemic-induced digitalization. Recently, the focus of the special questions has broadened and covered a greater variety of topics. This includes but is not limited to climate change, digitalization, minimum wages, supply chain issues, energy prices etc. The survey design makes it possible to quickly react to unforeseen and sudden events. Table 2 contains a detailed overview of special questions covered in each wave. The questionnaires for the individual waves contain more detailed information and are available for download from the project website: https://www.bundesbank.de/en/bundesbank/research/survey-on-firms. Additionally, the questionnaires contain questions on the firm characteristics as well as firm feedback. The data contains information on the firm’s self-reported employment and turnover as well as region and economic sector. Respondents also categorize their position within the firm and the firm type in general. Eventually they evaluate the perceived difficulty and length of the questionnaire. Finally, the BOP-F dataset also includes paradata, for instance, whether the interview was interrupted and the length of the interview. Table :Overview of topics in the special modules of BOP-F. Wave Questionnaire Topic(s) Covid- production and employment reactions, KfW credits, liquidity Short time work, VAT reduction, fintech, deferrals Climate change, inflation expectation, unemployment rate Liquidity sources, price changes, VAT effect Digitalization, quantitative changes Payment instruments, production costs, inflation expectations, negative interest rate, RD investments Second order inflation expectations, financing during financial crisis, Covid- & homeoffice, Blockchain State aid, investments, energy costs Russia–Ukraine war, price changes, corporate ratings, investment expenditure on software and hardware, energy prices/consumption, production costs and prices Russia–Ukraine war, safeguard clauses, minimum wage, climate policy measures Russia–Ukraine war, hiring plans, remaining debt and funding costs The “Bundesbank Online Panel –Firms”273
4 Data Access and Use for Research Access to BOP-F microdata is provided at the RDSC’s guest researcher workstations in a secure environment at one of the RDSC’s locations. This form of access is necessary in order to meet the statutory provisions concerned with safeguarding the confidentiality of statistical reports and, at the same time, to enable access to individual data for independent academic research purposes. To get access to the data, researchers are asked to submit a research data request including a research proposal. A research proposal is checked for feasibility of the research project given the research data, i.e. the suitability of the data to answer the research questions raised by the proposal. The research project must be of public interest. Commercial projects will not be accepted. One of the major advantages of BOP-F data is that it can be combined with other administrative Bundesbank data, which makes it particularly valuable for research. This includes but is not limited to other firm-level data such as balance sheet information (e.g. Janis) or information on foreign direct investments (e.g. MiDi). The data completes a series of Bundesbank survey datasets, including household surveys such as Bundesbank Online Panel –Households (“BOP-HH”) 9 or the Panel on Household Finances (“PHF”). 10 BOP-HH is also a high-frequency monthly survey. The structure and partly also the content of the questions are similar (e.g. with regard to inflation expectations), which allows a certain comparability between the company and household surveys. Although the data has not been available for long, there are already first papers that use the data. For instance, combining the survey data with administrative data, Boddin et al. (2020) investigate whether government-subsidized loans are an important policy measure to sustain the real economy during nonfinancially-driven crises like the Covid-19 crisis or whether they might in fact hurt long-run economic growth by keeping unprofitable firms alive. 11 Gärtner and 9For further information see https://www.bundesbank.de/en/bundesbank/research/rdsc/researchdata/bop-hh-757542. 10 For further information see https://www.bundesbank.de/en/bundesbank/research/rdsc/researchdata/phf-755932. 11 They find that firms identify pessimistic and uncertain beliefs about future demand, rather than current or future credit constraints, as the main impediment for their business. Demand-driven uncertainty is also what limits firms’s demand for credit. Firms predominantly rely on retained earnings to finance their operations. Those who access external financing do so largely through regular commercial loans rather than loans guaranteed by government programs. Firms that apply for government-guaranteed loans are more likely to display zombie features. The survival of these firms could hamper structural changes in the economy, which would worsen rather than improve the economic outlook in the medium to long term. 274 D. Boddin and M. Köhler