Contribution
The infant-industry argument is old and its evidence weak, because protection is chosen by the policy-maker for the whole country, so there is no counterfactual and no way to separate the mechanism from the implementation (p. 2). Here protection arrived without the policy-maker: the blockade displaced British trade onto circuitous routes, raising the cost of reaching northern France far more than southern France. Comparing departments by the size of that shock, the paper estimates the short-run effect of protection on adopting mechanized spinning and, using the shock as an instrument, the long-run effect of having more spinning when protection ended. It calls itself the first “well-identified, reduced-form evidence of an infant industry mechanism” (p. 6).
Key insight
A shock to trade costs that varies across regions within one country acts like regionally differentiated tariffs. If protection only shelters an industry, the advantage should vanish once trade reopens. If it lets the industry reach scale where learning by doing or agglomeration lowers costs, the protected locations should keep their lead after protection ends. They did, for the industry, for most of the century.
Assumptions
- The trade-cost shock is exogenous. The change in least-cost distance to London between pre-blockade routes and the blockade routes via Helgoland, Gothenburg, Gibraltar and Malta, with 1 sea km set equal to 0.15 land km (p. 20). Direct smuggling is excluded, on historians’ evidence that it was riskier (p. 21).
- No contemporaneous shock correlated with it (p. 26). Tested by pre-trends, placebo industries and controls for streams, coal, market potential, university access and literacy, each interacted with the treatment period (pp. 28–32).
- Exclusion restriction for the long run. The shock affects 1840–2000 outcomes only through 1812 spinning capacity (p. 34). The reduced forms are reported so that a reader need only assume the shock is uncorrelated with other determinants (p. 35).
- The sample. 88 of 109 departments observed in 1803 and 1812 (p. 25), fewer in the long run because France lost its territorial gains: 67–75 departments for persistence, 35 departments and 405 firms for productivity, 70–72 for value added (Tables 3–5).
- Spindles per thousand inhabitants, population held at 1811 (p. 25); 1803 spindles partly imputed by predictive mean matching (p. 21).
Key results
- Short run (Table 1, p. 27). Coefficient on log effective distance 33.11 (s.e. 9.78), stable at 24–41 with controls. Moving from the 25th to the 75th percentile of the shock predicts an increase in spinning capacity about equal to mean capacity in 1812. Empire-wide spindles rose about 370 per cent, from 380,000 to around 1.4 million (p. 22).
- Falsification (Table 2, pp. 30–32). Pre-trend 1794–1803 small and insignificant once market potential is controlled; no effect on capital-labour ratios, machine vintage, wool spinning or leather tanning. Most new capacity came from firm entry (p. 30).
- Persistence (Table 3, p. 36). One more spindle in 1812 from protection meant about 2–3 more in 1840 and 5–6 more in 1887 (2SLS). Southern departments shrank in absolute terms, which the author reads as the strongest evidence against slow diffusion from Britain (p. 36).
- Productivity (Table 4, p. 37). A one standard deviation increase in 1812 capacity meant firm labour productivity 0.2–0.5 standard deviations higher in 1840.
- Industrial value added (Table 5, p. 38). Positive and significant in 1860 (2SLS 0.00714); “The point estimate decreases in size over time and from 1896 onwards, the coefficient is longer distinguishable statistically from zero” (sic, p. 38). OLS is large and stable through 2000. No effect on agriculture or services.
- Weak instrument, conceded (p. 35). “The Kleibergen-Papp F-statistic for the first stage is generally below 10” KP F is 3.0–8.0 in Tables 3–5. The author argues that just-identified IV is approximately median unbiased and that the standard errors are not uselessly wide.
- Exports (p. 39). France went from net importer to net exporter of cotton goods; by 1828 cottons were 7.5 per cent of French exports. As late as 1850 France and Belgium spun more than other Continental countries.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Departments more shielded from British trade by the blockade added more mechanized spinning capacity, 1803–1812 | strong for this setting | DD with continuous treatment, placebos; Tables 1–2 |
| C2 | The protected departments kept and extended their lead in spinning through 1887 | moderate: weak instrument | 2SLS and reduced form, Table 3; KP F 3–8 |
| C3 | Spinning firms in those departments were more productive in 1840 | moderate | Table 4; output per worker, trimmed |
| C4 | Industrial value added per capita was higher in 1860 and not detectably so from 1896 | moderate | Table 5 |
| C5 | The mechanism is learning-by-doing externalities | weak: suggestive | firm entry, productivity, historical sources; p. 8 |
| C6 | France became internationally competitive in cottons | weak to moderate | aggregate exports, appendix figures not seen; p. 39 |
Concepts
- Trade-cost shock: the log change in each department’s shortest effective route to London between pre-blockade and blockade years (p. 20).
- Temporary protection: here, the period of the wars and blockade, 1803–1815. It was not a tariff, and it was not lifted into free trade: after the wars “tariff and non-tariff trade barriers were put in place between the two countries” (p. 33).
- Infant-industry mechanism: protection renders a location profitable in the long run, after protection is removed, through agglomeration economies (p. 33).
Connections
It belongs with the work on temporary shocks that permanently move economic activity (Davis and Weinstein 2002; Redding and Sturm 2008; Kline and Moretti 2014, p. 7), and with the infant-industry simulations it says cannot address the endogeneity of industry choice (Baldwin and Krugman; Head 1994; Irwin 2000). It sides with Crouzet against Heckscher, for whom the blockade’s industries were “hothouse” growth (p. 8).
Bearing on the record
- CLAIM-445 (protective regimes become self-maintaining). The claim
cites Juhász for its contrast between capability-building and
dependence-building protection: protection “need not be self-defeating”.
The reading supports that use and corrects the frame around it.
- Supports. Protection built a capability that outlasted it: the industry’s location persisted and spinning firms were more productive. That is a temporary measure leaving a lasting structure, the claim’s contrast case.
- Narrows. What persisted was the industry, not the protection. The paper says nothing about whether protection maintained itself, which is the claim’s mechanism. The aggregate gain faded: value added was higher in 1860 and not detectably so from 1896. The long-run estimates rest on a weak instrument.
- Corrects the frame. This was not “A temporary tariff” (the claim’s words). It was a war blockade, imposed against Britain and not to shelter cotton: “Napoleon remained highly ambivalent of developments in the cotton industry” (p. 24) and taxed raw cotton. Nor did protection simply end: post-war tariffs and non-tariff barriers followed (p. 33). So the within-France results show the protected regions’ lead under continued national protection; only the export evidence speaks to competitiveness at world prices, and the author says the within-country results do not show it (p. 6).
- A detail for the claim’s political-institutional feedback. The constituency formed. Footnote 7 cites a petition “from large spinners across the Empire requesting a complete ban on English cloth” (p. 13). The paper does not follow it up, and it does not say whether post-war tariffs owed anything to the new industry. That would be the test of whether this capability-building protection also built its own constituency.
- Filed as THEORY-919.
- CLAIM-446. Not touched. Nothing here is about evidence or anxiety.
- THEORY-885 (increasing returns). A historical case of a temporary shock locked in by increasing returns, at the level of industry location. It fits the account; it was not designed to test it.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- A draft, not the version of record. The AER paper (2018) may differ. Its abstract, which the LIT summary follows, puts the gain in industrial value added up to the second half of the nineteenth century and not later, which matches Table 5 here.
- Appendices not seen. The robustness tables, the trade and price series, the export and Continental comparison figures are in Online Appendices 1–3.
- Weak instrument in every long-run first stage.
- Small samples: 35 departments for productivity.
- Mechanism not identified. The author separates learning-by-doing from pecuniary externalities only suggestively, and says this “does not imply a case for infant industry protection in similar settings” (p. 8); where the prerequisites are absent, “infant industry protection can turn out to be an extremely blunt tool, irrespective of whether the policy would be optimal or not” (p. 40).
- Reporting slips. “is longer distinguishable” for “is no longer” (p. 38); “technlogy” in a heading (p. 11); the conclusion’s “the economy more generally” (p. 40) outruns Table 5, whose effect on value added is confined to 1860.
Open questions
- Did the spinners’ constituency shape the post-war tariffs? Archival work on the 1814–1816 tariff debates would show whether this protection maintained itself politically while it built capability.
- Does the version of record strengthen the first stage or change the value added results?
- Would a setting without post-war protection show the same persistence?