Recognition: unknown
Routine Work, Firm Boundaries, and the Rise of Local Supplier Entry
Pith reviewed 2026-05-10 00:25 UTC · model grok-4.3
The pith
Higher routine employment shares lead to more business applications through local supplier outsourcing.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
Across 722 commuting zones, a one percentage-point higher baseline routine employment share raises applications by 27.8 per 100,000 residents. Realized entry concentrates in micro-establishments, with no startup quality gains. Contract and industry evidence point to local supplier entry, not routine-manual displacement.
What carries the argument
Baseline routine employment share as an indicator of potential outsourcing demand that, for place-bound routine-cognitive tasks, spurs local supplier business applications.
If this is right
- Applications rise with higher shares of routine jobs in a local area.
- New businesses are small-scale micro-establishments rather than larger firms.
- No evidence of improved quality among the new startups.
- Patterns in contracts and industries confirm local supplier creation from outsourcing.
- Part of the overall increase in applications and decline in employer firms is attributable to boundary changes.
Where Pith is reading between the lines
- If outsourcing of routine work continues, local areas may see sustained growth in small supplier firms.
- Broader economic trends like automation could interact with this by either reducing or shifting such opportunities.
- Policy attention to business formation might consider support for supplier networks in routine-heavy sectors.
Load-bearing premise
The observed association between routine employment and applications primarily reflects demand for outsourcing to local suppliers rather than other confounding local economic conditions or trends.
What would settle it
Finding no rise in applications in commuting zones with high routine shares when outsourcing occurs to non-local providers, or lacking supporting contract evidence, would undermine the local supplier entry explanation.
read the original abstract
Between 2005 and 2019, U.S. business applications rose 40 percent while conversion to employer firms fell by nearly half. We study whether boundary redrawing helps explain this pattern. Structured routine-cognitive work can be governed through deliverables and thinner buyer and supplier interfaces. When such work remains place-bound, outsourcing creates demand for domestic specialist suppliers. Across 722 commuting zones, a one percentage-point higher baseline routine employment share raises applications by 27.8 per 100,000 residents. Realized entry concentrates in micro-establishments, with no startup quality gains. Contract and industry evidence point to local supplier entry, not routine-manual displacement.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper investigates the substantial rise in U.S. business applications between 2005 and 2019, coupled with a sharp decline in the rate at which these applications convert into employer firms. The authors argue that this trend is driven by firms redrawing their boundaries through the outsourcing of structured routine-cognitive work to local specialist suppliers, particularly when such tasks are place-bound. Using data across 722 commuting zones, they find that a one percentage-point higher baseline routine employment share is associated with 27.8 additional business applications per 100,000 residents. Realized entry is concentrated in micro-establishments, with no evidence of improved startup quality. Supplementary contract and industry evidence supports the local supplier entry mechanism over alternatives like routine-manual displacement.
Significance. If the central association can be credibly attributed to the outsourcing channel, this work advances understanding of how changes in task routinization and firm organization influence entrepreneurship and the structure of local economies. It provides a potential explanation for the observed increase in non-employer business activity and the stagnation in employer firm formation. The paper's strengths lie in its use of granular commuting zone-level variation, the focus on micro-establishment entry, and the integration of quantitative results with qualitative contract and industry evidence. This multi-pronged approach helps build a coherent narrative around boundary redrawing.
major comments (2)
- The main commuting-zone regression (Table 1 or equivalent in Section 4) reports a coefficient of 27.8 but provides insufficient detail on the full control set, including industry composition, prior CZ growth trends, or skill distributions. Without these, the estimate cannot isolate outsourcing demand from unobserved local confounders that may jointly determine routine shares and application rates.
- Section 3 (empirical strategy): the manuscript does not present an explicit identification argument, robustness checks, or alternative specifications (e.g., CZ fixed effects, lagged controls, or instruments) to support the claim that baseline routine employment share is conditionally exogenous to other factors raising applications. This is load-bearing for the boundary-redrawing interpretation.
minor comments (2)
- The abstract refers to 'contract and industry evidence' without summarizing its content or how it rules out routine-manual displacement; a brief parenthetical or footnote would improve clarity.
- Variable construction for 'routine employment share' and the exact baseline year should be stated more explicitly in the main text or an early appendix to facilitate replication.
Simulated Author's Rebuttal
We thank the referee for the thoughtful and constructive report. The comments highlight important areas for strengthening the empirical strategy and transparency of our results. We address each major comment below and commit to revisions that improve the isolation of the outsourcing channel while preserving the paper's core contribution.
read point-by-point responses
-
Referee: The main commuting-zone regression (Table 1 or equivalent in Section 4) reports a coefficient of 27.8 but provides insufficient detail on the full control set, including industry composition, prior CZ growth trends, or skill distributions. Without these, the estimate cannot isolate outsourcing demand from unobserved local confounders that may jointly determine routine shares and application rates.
Authors: We agree that fuller documentation and additional controls are warranted. The baseline specification in Table 1 already includes CZ population, the share of college-educated residents, and initial (2005) employment shares at the 2-digit NAICS level to capture broad industry composition. In the revision we will (i) report the complete coefficient vector and variable definitions in an expanded table, (ii) add lagged CZ employment growth rates for 2000–2005 as a direct control for prior trends, and (iii) incorporate finer occupational skill measures (e.g., routine-task intensity and education requirements) drawn from the 2000 Census and O*NET. These additions will be presented alongside the original results to demonstrate robustness. revision: yes
-
Referee: Section 3 (empirical strategy): the manuscript does not present an explicit identification argument, robustness checks, or alternative specifications (e.g., CZ fixed effects, lagged controls, or instruments) to support the claim that baseline routine employment share is conditionally exogenous to other factors raising applications. This is load-bearing for the boundary-redrawing interpretation.
Authors: We accept that an explicit identification discussion is needed. The routine-employment share is measured in 2005—prior to the acceleration in business applications—and is shaped primarily by long-run industry composition that is slow-moving and plausibly exogenous to post-2005 application surges conditional on initial industry and skill controls. In the revision we will add a dedicated identification subsection that formalizes this timing argument, report specifications that include lagged controls and alternative measures of routine intensity, and explore a historical-industry-mix instrument (following the spirit of Autor et al. and others) while acknowledging its limitations at the CZ level. We will also present results with state fixed effects as a robustness check. These changes will be clearly flagged as addressing the referee’s concern. revision: partial
Circularity Check
No circularity: central result is an independent cross-sectional regression coefficient.
full rationale
The paper's load-bearing claim is an OLS-style estimate relating pre-period routine employment share to subsequent business applications per capita across commuting zones. This coefficient is not derived from or reduced to any prior fitted parameter, self-citation chain, or definitional identity within the paper; it is a direct statistical association from observational data. No equations, uniqueness theorems, or ansatzes are invoked that would make the reported 27.8 applications figure tautological with the inputs. The analysis therefore contains no self-definitional, fitted-input-called-prediction, or self-citation-load-bearing steps.
Axiom & Free-Parameter Ledger
free parameters (1)
- routine employment share coefficient =
27.8
axioms (2)
- domain assumption Routine-cognitive work is sufficiently place-bound and contractible to generate local supplier demand when outsourced
- domain assumption Baseline routine employment share is exogenous to subsequent application rates conditional on included controls
Reference graph
Works this paper leans on
-
[1]
Skills, Tasks and Technologies: Implications for Employment and Earnings
Acemoglu, Daron, and David Autor.2011. “Skills, Tasks and Technologies: Implications for Employment and Earnings.” InHandbook of Labor Economics. Vol. 4, 1043–1171. Elsevier. Acemoglu, Daron, and Pascual Restrepo.2020. “Robots and Jobs: Evidence from US Labor Markets.”Journal of Political Economy, 128(6): 2188–2244. Ad˜ao, Rodrigo, Michal Koles´ar, and Ed...
2011
-
[2]
The Local Origins of Business Formation: Entry as a Two-Stage Process
19 Dinlersoz, Emin, Timothy Dunne, John Haltiwanger, and Veronika Penciakova.2026. “The Local Origins of Business Formation: Entry as a Two-Stage Process.” National Bureau of Economic Research w34881, Cambridge, MA. Dube, Arindrajit, and Ethan Kaplan.2010. “Does Outsourcing Reduce Wages in the Low-Wage Service Occupations? Evidence from Janitors and Guard...
2026
-
[3]
The sample consists of 722 commuting zones over a 15-year period, resulting in 10,830 commuting zone-year observations. 21 Figure 2 Year-by-Year IV Estimates of the Effect of RSH on Business Applications This figure plots 2SLS estimates of the effect of 2005 Routine Employment Share (RSH) on the cumulative change in business applications (per 100,000 resi...
2005
-
[4]
For years after 2005, the outcome is the change in BDS establishment entry between year 𝑡 and 2005; for years before 2005, the outcome is the backward cumulative difference between year 𝑡 and
2005
-
[5]
The 2005 coefficient is normalized to zero by construction
𝑅𝑆𝐻 𝑐,2005 is instrumented with the 1990 shift-share instrument, and vertical bars show 95% confidence intervals based on standard errors clustered by state. The 2005 coefficient is normalized to zero by construction. Because the pre-2005 points are backward cumulative differences relative to 2005 rather than formal event-study leads, we interpret them as...
2005
-
[6]
The 50+ category is constructed residually from County Business Patterns (CBP) total establishment counts
The size bins are 1–4, 5–19, 20–49, and 50+ employees. The 50+ category is constructed residually from County Business Patterns (CBP) total establishment counts. All specifications include state fixed effects and baseline commuting-zone controls. Standard errors clustered by state in parentheses. Observations weighted by commuting-zone population share. B...
2005
-
[7]
All specifications include state fixed effects and baseline commuting-zone controls
Columns (2)–(4) report estimates for changes between 2005 and 2016 in the Startup Formation Rate (SFR), the Entrepreneurial Quality Index (EQI), and realized entrepreneurial capital per input (RECPI). All specifications include state fixed effects and baseline commuting-zone controls. Standard errors clustered by state in parentheses. Observations weighte...
2005
-
[8]
Darker shades indicate greater exposure
RSH is defined as the employment share (percentage points) in occupations in the top tercile of the Deming (2017) routine task intensity distribution. Darker shades indicate greater exposure. 30 Figure A2 Geographic Variation in Business Application Growth, 2005–2019 Notes:This figure maps the change in business applications per 100,000 residents across U...
2017
-
[9]
Red shades indicate larger increases in business applications, while blue shades indicate declines. 31 Figure A3 Leave-One-State-Out IV Estimates Each point reports the 2SLS coefficient on RSH from the baseline specification after excluding all commuting zones in the indicated state. Horizontal lines show 95% confidence intervals based on state-clustered ...
1990
-
[10]
For years after 2005, the outcome is the change in BDS establishment entry between year 𝑡 and 2005; for years before 2005, the outcome is the backward cumulative difference between year𝑡 and
2005
-
[11]
The 2005 coefficient is normalized to zero by construction
Vertical bars show 95% confidence intervals based on standard errors clustered by state. The 2005 coefficient is normalized to zero by construction. As in Figure 3, the pre-2005 points are descriptive timing evidence rather than formal event-study leads. 33 Figure A5 Five-Year Timing Estimates for BDS Establishment Entry This figure plots 2SLS estimates o...
2005
-
[12]
𝑅𝑆𝐻 𝑐,2005 is instrumented with the 1990 shift-share instrument, and vertical bars show 95% confidence intervals based on standard errors clustered by state. Because these exercises rely on a different outcome measure than the baseline BFS regressions and apply common baseline controls across windows, we interpret them as supportive timing and placebo evi...
2005
-
[13]
Share of |ˆ𝛼𝑘 |
𝑎 Deming (2017). 𝑏 Acemoglu and Autor (2011). 𝑐 O*NET Work Context. 35 Table A2 Robustness Tests Panel A: 1980 Instrument (1) (2) (3) (4) RSH2005 ΔBus. App.sΔBus. App.sΔBus. App.s (First Stage) 2005–19 2005–10 2010–19 IV 0.709 ∗∗∗ (0.074) RSH2005 34.2∗∗∗ 7.2∗ 27.0∗∗∗ (6.7) (4.0) (6.9) Observations 722 722 722 722 Panel B: OLS (1) (2) (3) ΔBus. App.sΔBus. ...
2017
-
[14]
Panel A rebuilds the exact instrument using only the top-𝑘 industries ranked by absolute Rotemberg weight
The baseline row uses the exact leave-one-state-out instrument. Panel A rebuilds the exact instrument using only the top-𝑘 industries ranked by absolute Rotemberg weight. Panel B rebuilds the exact instrument after excluding the top-𝑘 industries. All specifications include state fixed effects and baseline controls, use CZ population weights, and report st...
2005
-
[15]
Top 5” and “Top 10
“Top 5” and “Top 10” refer to standardized 1990 CZ employment shares for the industries with the five and ten largest absolute Rotemberg weights. “Joint𝐹” reports the absorbed first-stage joint significance test from the regression of RSH on the listed shares. KP𝐹 reports the Kleibergen–Paap rk Wald𝐹-statistic. Hansen 𝐽 𝑝 -values come from the overidentif...
1990
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.