Panel entry and exit: separate growth from composition
A panel’s aggregate can change because continuing companies changed, because new companies entered or because old companies disappeared. Separate these components before calling an increase broad-based growth. A larger current total can coexist with declining revenue among every company that appears in both periods.
Distinguish an economic panel from a reporting boundary
A company’s reporting perimeter determines which businesses its statement includes. A research panel determines which observations your cross-company analysis includes. Even when each issuer’s reporting perimeter is stable, a vendor’s coverage expansion, a listing or a delisting can change the panel composition. Record both boundaries rather than assuming one check solves the other.
The US Census Bureau’s Business Dynamics Statistics FAQ discusses longitudinal business measurement and entry and exit concepts. It provides useful context for why consistent populations matter. The revenue bridge below is our illustrative research worksheet, not an official BDS statistic or a claim that listing events equal economic establishment births.
A hypothetical two-period bridge
Assume a panel contains issuers A, B and C in the first period, with revenue of 60, 40 and 20. In the second period, A has 54 and B has 36, C is absent, and newcomer D contributes 50. The published panel total rises from 120 to 140, or 16.7%. Yet the continuing issuers decline from 100 to 90, or 10%.
| Component | Change in panel revenue | Interpretation |
|---|---|---|
| Continuing A and B | -10 | Observed within-issuer decline |
| Entrant D | +50 | Added current-period observation |
| Exiting C | -20 | Removed prior-period observation |
| Total bridge | +20 | 140 minus 120 |
The identity is current total minus prior total equals continuing-company change plus current revenue of entrants minus prior revenue of exits. It is an accounting decomposition of this dataset. It does not tell you whether D grew economically from zero or whether C ceased operations. They entered and exited the panel; the underlying businesses may have existed throughout.
Classify why records enter or leave
Use separate reasons for new economic activity, listing changes, mergers, identifier changes, coverage additions and missing disclosure. A renamed issuer should not appear as one exit and one entrant if the economic entity is continuous. A merged company may need a carefully documented bridge rather than automatic survival or automatic disappearance.
Check whether absences are temporary. A delayed report can create an apparent exit in one data pull and reappear after the publication date. Preserve the extraction date and completion status. If you compare an incomplete current period with a fully populated prior period, the bridge may mostly measure reporting timing.
Report both views without overstating either
The full-panel total describes the observations actually included in each period. The continuing-company result describes the matched subset. Neither automatically measures the entire market. The matched panel can create survivorship bias by excluding failed companies and new businesses. Its useful role is isolating one component of change, not establishing the definitive market growth rate.
Show the number of issuers and revenue coverage in each group. If the continuing sample represents only half the earlier panel revenue, its growth rate needs that context. If entrants have missing prior values, do not fill them with zero unless the research question and evidence justify genuine absence of prior economic activity.
Audit the panel identity
Preserve stable entity identifiers, mapping decisions, period definitions and group membership. Reconcile the bridge to the two published totals, then trace a few entrants and exits back to filings. When the bridge fails to reconcile, investigate unit changes, duplicate joins and currency conversions before attributing the difference to economics.
An AI narrative should state which part of growth comes from continuing companies and which part comes from composition. In the example, saying the panel expanded while continuing issuers contracted is a precise observation. Saying the industry’s demand grew 16.7% goes beyond what this panel establishes.
FAQ
Does a new listing count as a new business?
Not necessarily. The company may have operated privately for years. Its entry into the research dataset is a different event from economic creation.
Is a matched panel always preferable?
No. It helps isolate within-company change but can omit failures and new activity. Present its coverage and the full-panel bridge together.