By Ian Hathaway
I recently read a paper that took an innovative approach to at least partially answer a question that is boggling the minds of many economists and other observers: why has the firm formation rate declined precipitously the last three decades? I think it’s one of the most important topics in the economics profession today, and warrants a great deal of continued research in the coming years.
Aside from liking what was new about this paper, something else stood out to me—something I’ve seen before. The paper itself isn’t important here, so I’m not going to reference it explicitly—I’m not one to needlessly criticize someone else’s hard work, particularly when doing so isn’t central to the argument I’m trying to make.
The paper analyzes the relationship between, let’s say variable X, and the “startup rate,” defined as the rate of new establishment formations, not firm formations. This was done, presumably, because data on the former are much easier to obtain. But firms and establishments are not the same, and evidence suggests it has increasingly become important to distinguish between the two.
In research published last year by Mark Schweitzer, Scott Shane, and myself, we showed that the source of new business establishments is increasingly coming from existing firms, or what we call “new outlets”:
This is something that is not occurring in isolation. The data show that this trend has happened in each broad industrial sector (9 SICs sectors) and across each state (data are not available at the metropolitan area level) during the last three and a half decades:
![]()
I recently read a paper that took an innovative approach to at least partially answer a question that is boggling the minds of many economists and other observers: why has the firm formation rate declined precipitously the last three decades? I think it’s one of the most important topics in the economics profession today, and warrants a great deal of continued research in the coming years.
Aside from liking what was new about this paper, something else stood out to me—something I’ve seen before. The paper itself isn’t important here, so I’m not going to reference it explicitly—I’m not one to needlessly criticize someone else’s hard work, particularly when doing so isn’t central to the argument I’m trying to make.
The paper analyzes the relationship between, let’s say variable X, and the “startup rate,” defined as the rate of new establishment formations, not firm formations. This was done, presumably, because data on the former are much easier to obtain. But firms and establishments are not the same, and evidence suggests it has increasingly become important to distinguish between the two.
In research published last year by Mark Schweitzer, Scott Shane, and myself, we showed that the source of new business establishments is increasingly coming from existing firms, or what we call “new outlets”:
Figure 1: Distribution of New Establishments by Type (1978-2012)
 |
| Source: Census Bureau, BDS; author's calculations Click for larger image |
Furthermore, because new outlets are generally larger than are new firms in terms of employment, economic activity at new business establishments (as measured by employment) occurs in no small part in new outlets:
Figure 2: Distribution of Employment at New Establishments (1978-2012)
 |
| Source: Census Bureau, BDS; author's calculations Click for larger image |
This is something that is not occurring in isolation. The data show that this trend has happened in each broad industrial sector (9 SICs sectors) and across each state (data are not available at the metropolitan area level) during the last three and a half decades:
Figure 3: Share of New Establishments that are new Firms by Sector (1978-2012)
 |
| Source: Census Bureau, BDS; Author's calculations Click for larger image |
Figure 4: Ratio of New Firm Share of New Establishments in 2012 over 1978 by State
 |
| Source: Census Bureau, BDS; author's calculations Click for larger image |
Among the sectors, the shift from new firms to new outlets as the source of new business establishments was greatest in finance, insurance, and real estate, the very broad transportation and utilities (which includes communications) sector, and in construction. Agriculture, retail trade, and mining saw the smallest changes over the 34-year period—but changed nonetheless.
For the states, the chart shows the 2012 new firm share of new establishments over that same share in 1978. Since the data are sorted by largest ratio at the top, states experiencing the smallest shifts from new firms to new outlets as the source of new business establishments are at the top, while those witnessing the most change are at the bottom.
In short, the importance of new outlets to the formation of new establishments has grown steadily and significantly during the last few decades, and this shift has occurred in each of the broad industry sectors and US states. As a result, studies that substitute new establishments for new firms—including when exploiting industrial and geographic differences—are increasingly using less precise estimates over time.
Ian Hathaway is a Nonresident Senior Fellow at the Brookings Institution. You can follow him on Twitter @IanHathaway.