This article was originally written for the OECD Statistics blog
The business sector contributes 63% to GDP in the OECD (in 2022) and is a key driver of economic development, trade, innovation, and job creation. It is also a key source of emissions. OECD’s business statistics are therefore an important tool for monitoring economic growth and the transition to climate neutrality. In addition, to reduce their carbon footprint, firms must innovate with cleantech and digital solutions. This means radical changes in how they produce, invest and trade, as well as a greater need to monitor these changes. The digitalisation that occurs during these processes presents opportunities for new high-quality data to complement traditional business statistics, with the additional benefit of mitigating the survey fatigue experienced by firms in OECD countries. This blog deals with statistical issues that are particularly relevant to Small and Medium-sized Enterprises (SMEs), but many of them also apply to business statistics in a broader sense.
Business data do not grow on trees
With increasing globalisation, digitalisation and climate pressures, users are demanding broader and timelier measures of businesses’ activities. Traditionally, National Statistical Offices (NSOs) have been the primary producers of official statistics. Certain International Organisations (IOs), including the OECD, contribute to data production through technical assistance and the development of statistical standards and guidelines.
Business demography, firm births and deaths, and changes in size, are tracked through registration and liquidation procedures, or social security and tax declarations. This information is later compiled into national business registers and used for national accounts. OECD then use these to feed databases, like the “OECD Structural and Demographic Business Statistics” or the “Timely Indicators of Entrepreneurship”. Further insights come from financial reporting, or targeted surveys and economic censuses that underlie statistics on firm productivity and performance. Similar approaches are replicated at a subnational level to measure regional disparities in business dynamics and performance and inform regional development and cohesion policies. Data coverage is usually dictated by residency and the regulatory frameworks under which firms operate, e.g., legal definitions and administrative thresholds, affecting international comparability and the efforts required to close gaps.
From data collection to integration and publication, hundreds of experts are involved worldwide. The process can take months, resulting in sometimes significant delays before data become available for policy making.
Business surveys have been effective in gaining more granular insights more quickly
Business surveys offer a solution to the lack of granularity and timeliness. During the COVID-19 pandemic, when usual statistical systems could not follow developments fast enough, they were used extensively to assess the impact of lockdowns and market disruptions on SMEs and entrepreneurs.
Business surveys are conducted in very different contexts with a wide variety of scopes and ambitions. Some have since been mainstreamed and are now fully integrated into the business statistics landscape. Run on a regular basis, often by NSOs or IOs, they monitor business activities, firm-level engagement in research and development, innovation, or digitalisation. They gather firms’ views on the business conditions they face, including corruption, informality, competition barriers, regulation, or taxation etc. They can also be complemented with labour force, adult or employee surveys to gain for example insights into questions around human capital and skills.
However, as with any survey type, there are issues of sampling and representativeness to consider. By design or by uptake, surveys can be biased towards certain firms, distorting the messages conveyed to policy makers. Typically, micro-firms with less than 10 employees are not included in key regular data collections, meaning over 90% of the OECD business population may be obscured. Some sectors like agriculture and network industries might be omitted as well. And too small sample sizes reduce the level of granularity and the benefits these surveys can offer. These problems are exacerbated by declining response rates in many OECD countries.
Microdata, the Holy Grail or a stopgap solution?
First and foremost, policy makers need evidence of causes and effects, i.e. what drives changes in business conduct and business performance. Firm-level microdata are the key to better understanding the effects. They reflect the characteristics and trajectories of individual firms, enabling disaggregation and granular economic analysis. Microdata comes from business registers and administrative sources, or are created via surveys and censuses, directly by the NSOs, private or non-profit institutions. Examples include patent offices through the filing of intellectual property applications, or business analytics companies which compile information on corporate financial situation (ORBIS Moody’s Analytics), cross-border greenfield investments (Financial Times fDi Markets), or startups development (Crunchbase). As part of the project on Unleashing SME potential to scale up, the OECD has worked with the European Commission and NSOs to leverage this wealth of information, and develop evidence on high-growth firms. This pilot data shows that firms of any size and location can scale up. If young high-tech ones are more likely to, the typical scaler is a mature SME in a low-knowledge-intensive sector, revealing opportunities beyond start-ups. Possibly even more important, about 60% of scalers sustain their new scale over time, and 26%-35% of them manage to scale up again. This indicates a transformation grounded on durable productivity gains and implies that scale up policies can pay off.
That said, the use of microdata can be constrained by challenges associated with representativeness, quality control and confidentiality, in fact very similar to those of administrative data. Microdata may lead to conclusions that are inconsistent with macrodata trends or over time, e.g. from one survey round to another. Validation methods are imperfect and rely on strong assumptions. Regional microdata is also scarce, of uneven quality, and can fall short in informing policies. Furthermore, access to official microdata is protected, often by law, with measures implemented to prevent identification or disclosure. It can typically only be used under close supervision, or through arrangements preventing direct access, e.g. via remote facilities or exchanges of algorithms and results.
Pushing the data frontier
Representativeness and timeliness are both crucial and notably difficult to achieve in business statistics. In principle, surveys can address these issues, but they rely on the goodwill of respondents. Declines in response rates and quality call for new solutions to be found, with opportunities to explore in emerging technologies, alternative and complementary data sources, and by combining approaches. Examples follow.
Digitalisation creates new possibilities to monitor business. More digital firms means more data. But while the digital transformation affects all business functions and sectors, not all enterprises progress equally fast. Size matters for data scale. SMEs often lack a data culture and the volume of data required for machine learning and advanced analytics. This means that more time will be needed before new data generated by digital processes in smaller firms will be fully useful for statistical purposes.
Satellite data potentially offers economic insights on firm performance such as expansion and shifts in production through space imagery that tracks supply chain logistics, physical assets (e.g. vehicle, crops, and equipment), and environmental impact (such as air quality, water pollution, land erosion). The full potential of geospatial analysis would appear large, although it hinges on high-resolution imagery and the capability to geo-reference data and link location-based data to business characteristics. Linking data sources is also hugely promising, including the linking of data from new sources with official business registers and/or administrative data. The OECD’s Trade by Enterprise Characteristics database combines cross-border trade data with business registers to measure the export performance of different businesses. These data shed light on SME participation in international trade. Recent OECD research employs machine learning using information from one million SME websites to gauge their environmental engagement, and link results with commercial financial data on firms (ORBIS) to identify greening SMEs.
Finally, the climate crisis calls for green practices, regulation initiatives and business adaptation. Firms are increasingly aware of the importance of including environmental considerations and natural capital in their reporting, and of complying with responsible business standards. This is the right time to reflect collectively on how this relates to official business statistics. In that direction, the statistical community is working to align business accounting with the System of Environmental Economic Accounting, an international standard similar to the System of National Accounts, for integrating environmental and economic statistics.
Co-investment, cooperation and continuity for more and better data
Meeting the policy demand with Smart Data innovation requires pooling resources, investing in data commons, maintaining continuity in production, and ensuring data are both high-quality and fit for purpose.
Given the magnitude of the task, data actors, both in the public and private sectors, can and should join forces. The Wiesbaden Group has long discussed the adaptation of business registers to policy needs. The Development Data Partnership, a global consortium between private digital platforms and IOs, has established common legal and technical frameworks for shared access to timely, privately-held data, including on micro-firms. There are also key technical advancements, like the definition of Statistical Data and Metadata Exchange (SDMX) standards and systematic scraping of open data through Application Programming Interfaces (APIs) increasingly offered by data providers, reflecting heightened cooperation, even beyond business statistics.
The permanence of data cooperation is vital for consistent data production, data affordability, and policy relevance. This continuity, especially with the private sector, requires an alignment in vision, and clearly defined terms of cooperation. This can be achieved through long-term win-win arrangements, or by means of regulation. Indeed, regulation may be needed to prevent private data owners from changing data access terms and pricing unilaterally. This can be the case in markets for specific data where production relies on one or a few producers whose position is not contestable due to lack of competitors.
The next frontier consists in fostering data creativity and skills, and scaling up digital infrastructure for storage, processing, preservation of confidentiality and security. Also, statistical quality frameworks need to be modernised to ensure that statistics based on new approaches are reliable and fit for use.
Business as usual will no longer suffice for creating the business data we need. We must advance further to ensure that our data effectively supports governments and companies in taking the right actions.
Rudiger Ahrend is Head of the Economic Analysis, Data and Statistics Division in the OECD Centre for Entrepreneurship, SMEs, Regions and Cities. In addition, he also oversees the activities of the OECD Laboratory for Geospatial Analysis. In these capacities, and in his previous role as Head of the Urban Programme, he has been supervising numerous projects in a wide area of subjects, including on industrial transition, regional and urban innovation and development, subnational finance, spatial productivity, metropolitan governance, land use, land value capture, housing, green growth, climate change, transport, metropolitan governance, and national urban policies. He has also supervised numerous reviews and case studies of regions and major metropolitan agglomerations, and is the main author of “The Metropolitan Century: Understanding Urbanisation and its Consequences”. During his time as Head of the Urban Programme, Dr Ahrend was also in charge of the OECD Working Party on Urban Policies, as well as the OECD Roundtable of Mayors and Ministers.
Sandrine Kergroach is Head of SME and Entrepreneurship Performance, Policies and Mainstreaming unit at the OECD Centre for Entrepreneurship, SMEs, Regions and Cities (CFE). She leads the work on innovation, internationalisation and the scaling up of SMEs and start-ups, their productivity and ESG performance. She supervises activities related to policy monitoring, the development of data infrastructure and the OECD SME and Entrepreneurship Outlook. She also leads efforts for mainstreaming SME&E policy considerations. Sandrine holds a Doctorate in Economics (TU Berlin), a Master in Strategy and Management (Paris Dauphine-PSL), a Master in Modern History (Paris Sorbonne) and a Bachelor in Applied Economics and Statistics (Paris Dauphine-PSL).
Annabelle Mourougane is heading the Trade and Productivity Statistics Division in the OECD Statistics and Smart data Directorate. She holds Master Degrees in Economics and Econometrics (ENSAE) and in Macroeconomics (EHESS and École Polytechnique). She joined the OECD in 2001 and has held a number of senior positions in the Economics Department and in the Trade and Agriculture Directorate. Prior to joining the OECD, Ms Mourougane worked at the French National Statistical Agency (INSEE), the European Central Bank and at the French Fiscal Council/Court of Audit (in 2013-2015). She has published in international academic journals, in the areas of forecasting and modelling, trade, labour market and fiscal policy.