Readers will learn which indicators statisticians prioritize, how administrative sources shape coverage and timing, and why survey programs are a useful complement when registers omit non-employer or informal activity.
How entrepreneurship economic growth depends on official startup counts
Reliable counts of new firms are central when interpreting entrepreneurship economic growth because researchers link firm births to later employment and turnover to assess economic impact. The international guidance lists firm births, employer status, employment and turnover as core measurement elements, which shape what a statistic labeled a ‘startup’ actually captures OECD entrepreneurship indicators
Official series are measurement tools, not forecasting machines. Different datasets are designed to answer different questions: some show near-term application activity, while others report surviving employer firms over years. Users should therefore match the dataset to the question they have in mind.
Unit of analysis choices, such as whether a series counts establishments or enterprises and whether non-employer businesses are included, change how many and which startups appear in a count. These definitional differences mean that two official series from the same country can tell different stories about entrepreneurship economic growth.
Core indicators used by official statistics
Statisticians typically rely on four core indicators when counting startups: firm births, employer status, employment, and turnover. These indicators are named together in international metadata and guidance and form the backbone of harmonized entrepreneurship statistics Eurostat business demography guidance
1. Firm births: identifies newly formed legal units or businesses that meet a predefined entry criterion.
2. Employer status: distinguishes firms that employ others from sole proprietors without payroll, which affects how employment gains are counted.
3. Employment: measures jobs associated with a cohort of newborn firms and is essential for linking firm births to economic growth.
4. Turnover: tracks revenue flows and is used together with employment to identify high-growth firms and economic contribution.
Before comparing series, consult the metadata pages for precise operational definitions, such as the Eurostat business demography statistics page, because seemingly similar labels can rest on different rules and look-back windows.
Administrative sources explained: business registers, tax and social records
Most official startup counts are built from administrative sources: national business registers, tax filings, and social security or payroll records. These inputs determine coverage and timeliness in official statistics OECD entrepreneurship indicators
Business registers commonly record legal units and identifiers used to track births and deaths of firms. Tax records and social security data can provide employment and payroll information that helps determine whether a new registration corresponds to an employer firm.
Quick verification checklist for administrative startup series
Use before citing a series
There are frequent coverage gaps: late registration can mean a firm operates for months before appearing in a register, and informal or non-employer activity often does not appear at all. National office metadata and release notes usually explain how often registers are updated and when revisions are applied.
Tool: how to access administrative datasets and documentation
Start with the metadata and dataset landing pages maintained by international and national statistical bodies. OECD and Eurostat provide harmonized metadata, while national statistical offices publish dataset pages and documentation for each series OECD entrepreneurship indicators, and the OECD data explorer
Key access points include national business register portals, the dataset landing pages at statistical offices, and documentation sections that describe unit of analysis, look-back windows, and revision policies.
Checklist for evaluating a register:
- Unit of analysis: establishment or enterprise?
- Employer status: are non-employer firms included?
- Look-back window: how many months or years are used to define a birth?
- Revision policy: how are late updates handled and communicated?
When you need linked microdata or longitudinal files, check national office access policies and microdata request procedures; information on application processes and confidentiality safeguards is typically published alongside dataset descriptions. For assistance, see the contact page.
United States case: BFS versus BDS and what each counts
In the United States, two commonly cited series serve different purposes: the Business Formation Statistics (BFS) provide near-real-time counts of business applications, while the Business Dynamics Statistics (BDS) report historically consistent employer-firm births and survival over time. The Census Bureau describes these series and their intended uses BFS landing page
BFS tracks new business applications and can signal rapid changes in application activity, making it useful to spot short-term trends in entrepreneurship indicators. BDS, by contrast, constructs cohorts of employer firms and follows them through time to report entries, exits and net employment effects.
Because BFS measures application activity and BDS measures surviving employer-firm births, they should not be used interchangeably. Use BFS for timely application trends and BDS for long-term firm dynamics and employment contributions.
Survey complements: what GEM and other programs add
Survey-based programs such as the Global Entrepreneurship Monitor collect data on nascent entrepreneurship, motivations and informal activity that administrative registers may miss. These surveys therefore complement register-based counts for a fuller picture of entrepreneurship economic growth GEM global report
Surveys capture intentions to start a business, early-stage efforts, and reasons for entrepreneurship that are invisible in business registers. That makes surveys particularly useful when you want to understand the human or motivational side of startup activity rather than just registered counts.
Join Michael Carbonara's campaign updates
For immediate next reads, consult the OECD-Eurostat metadata pages and the U.S. BFS and BDS landing pages to match your research question with the right dataset.
Keep in mind that survey sampling, question wording and timing differ from administrative record collection, so combining these sources requires attention to coverage and comparability.
CTA: where to go next with startup data
Good first steps are reading OECD-Eurostat metadata, and then visiting the U.S. Census pages for BFS and BDS to see how the series are constructed and updated BDS program page. See the about page for author information.
After identifying a candidate dataset, check the unit of analysis and look-back window to ensure that counts answer the question you need. If you require deeper analysis, plan for requests for linked microdata or seek academic work that uses those linked files.
Common comparability problems across countries
Cross-country comparisons are difficult because countries differ in unit of analysis and whether non-employer firms are included. OECD and Eurostat harmonization efforts stress these limits and encourage checking country metadata before comparing numbers OECD entrepreneurship indicators
Examples of comparability issues include counting an establishment in one country versus an entire enterprise in another, or treating re-registrations and changes of ownership differently. These choices can produce large differences in reported startup counts even if economic activity is similar.
Before reporting a comparison, read the country-specific metadata and method notes to understand how each series defines a birth and handles administrative events like re-registrations.
Counting non-employer and informal startups: limits of registers
Administrative registers often miss non-employer firms and informal solo entrepreneurs because these activities do not generate the same filings as employer firms. Survey programs can reveal these gaps by reporting entrepreneurial activity that does not appear in registers Eurostat business demography guidance
How many new small businesses never show up in official registers?
Official statistics count startups using core indicators such as firm births, employer status, employment and turnover drawn from administrative registers and surveys; understanding the dataset's unit of analysis and metadata is essential to interpret what a startup count actually measures.
Indicators that suggest undercounting include divergence between high survey-reported entrepreneurship rates and low register-based birth counts. When that occurs, analysts should combine register measures with survey evidence to avoid misleading conclusions about entrepreneurship economic growth.
Measuring high-growth startups and gazelles: what the literature shows
Identifying high-growth firms commonly requires linking birth records with multi-year employment and turnover data so researchers can observe rapid job and revenue growth after firm birth. Foundational literature describes this approach and it remains a basis for modern firm dynamics work Survey on gazelles and job creation
Multi-year follow-up matters because short-term signals can misrepresent a firm’s long-term impact. Studies that label firms as high-growth typically require several years of linked employment and turnover observations to verify that a firm truly expanded at an above-average pace.
Practical steps to find and interpret startup data in 2026
Follow a short ordered checklist when you start with a dataset: identify the dataset landing page, check unit of analysis, read the look-back and revision notes, and then select the indicator that matches your question. OECD guidance is a useful reference for harmonized definitions OECD entrepreneurship indicators
- Identify dataset and landing page
- Check unit of analysis and whether non-employer firms are included
- Read look-back window and revision policy
- Pick indicators that match your research question
Prefer BFS-like application series when you want near-term trends, and prefer BDS-like cohort series when you need long-term firm dynamics and employment analysis. When in doubt, cite the dataset name and link to its documentation.
Examples and scenarios: interpreting conflicting signals
Scenario A: rising business applications but flat employer births. If business application counts rise quickly but employer births remain flat, plausible explanations include a lag between application and hire, increased non-employer activity, or more applications that do not convert to employer firms. Checking the BFS and later BDS cohorts can clarify the sequence of events BFS landing page
Scenario B: high survey entrepreneurship but low register counts. If survey measures show many people starting or intending to start firms but registers report few new births, this pattern may indicate high informal activity or delayed registration. Combining survey results such as GEM with register data helps resolve the discrepancy GEM global report
In both scenarios, consult metadata and consider requesting linked microdata or academic studies that trace cohorts over time to produce a fuller interpretation.
Typical mistakes and pitfalls when using startup counts
Frequent errors include treating BFS application counts as direct measures of employer-firm entries, failing to check the unit of analysis, and ignoring revision histories. The OECD guidance and national metadata can help prevent these mistakes OECD entrepreneurship indicators
Avoid overinterpreting short-term signals. Application spikes can reflect seasonality or reporting changes rather than sustained firm creation. Always label the dataset and note its intended measurement concept when publishing numbers.
How journalists, policymakers and researchers should report startup data
Good reporting requires naming the dataset, stating the unit of analysis and date, and linking to the primary documentation where possible. Attribution language such as ‘the dataset shows’ or ‘according to’ helps readers understand the source and limits of the number Eurostat business demography guidance
Recommended caveats include noting coverage limits, whether non-employer firms are included, and any recent revisions. If making claims about economic impact, link births to employment and turnover evidence or cite longitudinal studies that do so.
Data gaps and open issues: platform-based and informal entrepreneurship
Platform-based businesses and gig work pose measurement challenges because income flows can bypass traditional firm registries and payroll systems. The need for clearer metadata and new survey modules to capture these activities is an open issue in the literature GEM global report
Researchers and statistical offices are exploring how to align short-term application series with long-run survival statistics and how to expand metadata to support real-time policy use. Until metadata improve, users should combine registers with survey evidence when platform or informal activity is likely important.
Conclusion: reading entrepreneurship economic growth statistics responsibly
Key takeaways are straightforward: core indicators like firm births, employer status, employment and turnover matter; administrative sources determine coverage and timing; surveys complement registers by capturing nascent and informal activity. Check definitions and metadata before citing numbers OECD entrepreneurship indicators, and see the Eurostat entrepreneurship indicators. Also visit the Michael Carbonara homepage for related posts.
Final checklist before publishing: check dataset name and landing page, state the unit of analysis and date, and cite the primary documentation. These steps help ensure responsible use of startup counts when discussing entrepreneurship economic growth.
A firm birth is a new legal unit or business that meets a dataset's operational criteria; exact definitions vary by dataset and are detailed in metadata.
Use BFS for near-term signals in business application activity and BDS for long-term employer-firm births and survival analysis.
Compare register counts with survey estimates and check for divergences; read metadata on coverage and look-back rules to spot likely undercounts.
