Material effects of trade barriers

A favoured argument in defence of tariffs, quotas and other trade barriers is to shield emerging (or ‘infant’) industries from foreign competition during a period of establishment and early growth. By temporarily raising the cost of imports, governments can give an industry breathing space to build up a critical mass of capacity and talent, which would otherwise risk being snuffed out.

Critics would say that this infant industry motivation often falls short in practice: at what point can the ‘infant’ be considered to have grown up, such that trade barriers are no longer required? What if the ‘infant’ never really grows up?

Establishing the merits of trade barriers in an infant-industry context is inherently complicated, as it requires one to disentangle economic and political arguments. (If an emerging industry successfully lobbies for trade barriers to be imposed, why would its lobbying power be any less if that industry subsequently establishes itself as a major source of employment and economic activity?) To isolate the economic consequences, an ideal scenario for analysis would be trade barriers that were imposed — and later removed — for reasons totally separate from a particular infant industry. A war, for example.

This is precisely the starting point Juhász adopts. Within a difference-in-difference framework, she examines the effects of trade barriers introduced during the Napoleonic Wars on French cotton manufacturing. The upshot? While trade barriers were only imposed for a period of years, France’s cotton manufacturing industry was substantially changed, with effects observable decades later.

You shall not pass

The Napoleonic Wars took place between 1803 and 1815, with the French empire (under Napoleon) on one side, and the British empire on the other. The conflict had wide-reaching consequences across Europe, with even notionally neutral states swept up in events. Napoleon was able to exert influence across the bulk of continental Europe.

Napoléon Bonaparte by Jacques-Louis David. Source: Wikimedia Commons.

One example of this influence is the Continental Blockade: trade sanctions imposed on Britain between 1806 and 1813. The purpose was to limit Britain’s ability to sell goods in Europe, thereby weakening its economic power. The blockade was atypical in the sense that Napoleon did not attempt to stop trade in and out of British ports; rather, the blockade was imposed across European ports against vessels carrying British goods.

Napoleon’s blockade was not 100 per cent effective. British goods were still transported to and sold in continental Europe. Nevertheless, the blockade stopped the most direct trade routes with respect to France. Instead, goods were typically smuggled into Europe via four ports: Gothenburg and Helgoland in the north, and Gibraltar and Malta in the south.

Consequently, the blockade against British trade had more pronounced effects in the north of France relative to the south of France. To better understand this, consider the figure below, which plots the number of ships travelling from Britain to various European ports. Prior to the blockade, a substantial degree of UK–Europe trade activity was concentrated in the north of France and the Netherlands. This activity all but ended once the blockade was imposed.

Changes in trade patterns: number of ships traveling from Britain to European ports

Number of ships travelling from Britain to European ports. Source: Figure 1, Juhász (2018).

The net effect of this shift in trade activity is that the relative cost of importing British goods increased more strongly in the north of France than in the south. To estimate this, Juhász measures the change in effective trade distance for each département (administrative area) across France — from the nearest port pre-blockade to the closest of the four smuggling hubs during the blockade.

As two examples of this: the area of Seine-Maritime (on France’s northern coast) witnessed an increase in effective distance to London from 86 kilometres pre-blockade to 1282 kilometres during the blockade. Its nearest smuggling port was Helgoland. By contrast, the southern area of Pyrénées-Orientales started with an effective distance of 548 kilometres pre-blockade, which increased to just 689 kilometres under the blockade.

Pure spin

Juhász chooses to focus her analysis of the blockade’s effects on cotton manufacturing. The reason for this relates to the differences in British and French cotton production at the turn of the century.

At the outset of the war, Britain was a world leader in cotton textiles. It had pioneered a new technology — mechanised cotton spinning — which greatly increased industry productivity. This was among the first examples of industrialisation, bringing together mechanisation and factory-based production. Mechanised cotton spinning, once developed, spread rapidly through the British cotton industry. The relative low cost of the machines is a key factor: there were few direct barriers to their production, installation and use. What one needed was know-how.

Reinventing the wheel. Source: Wikimedia Commons.

By contrast, France’s cotton industry still largely relied on smaller-scale cotton spinning wheels. The new British technology was certainly known in France, but the ability for French manufacturers to adopt it was limited by British restrictions on the export of the technology and relevant know-how.

Inevitably, such technology restrictions could only slow — rather than stop — the spread of the industrial breakthrough. And here the infant industry argument comes to the fore: Juhász asserts that the learning curve associated with mechanised cotton spinning meant that France’s late adoption put it at a distinct competitive disadvantage to Britain. Widescale investment in the new technology was not commercially justified so long as cheaper British cotton could readily be imported.

The blockade thus provided the right settings to enable a technology — and geographic — shift in French cotton manufacturing.

How the wheel turns

Juhász plots the location of cotton manufacturing activity across France both before and toward the end of the blockade. As the maps below show, manufacturing activity was dotted across the country in 1803, including a major centre of production in and around Lyon toward the south. By 1812, there is a clear shift in production intensity: increasing in the north, declining in the south.

This illustrative evidence supports the hypothesis that Juhász advances. With direct trade with Britain essentially halted by the blockade, the trade cost shock had a greater effect on the north of France (geographically closest to Britain) relative to the south. In turn, the trade protection afforded by the blockade was more advantageous for northern manufacturers than their southern peers.

Short-run changes in cotton spinning capacity: 1803–1812

X denotes missing observations for one or both years. Source: Figure 3, Juhász 2018.

In an econometric examination of the short-run effects of the blockade, Juhász employs a variation of the difference-in-difference method. This approach, in broad brushstrokes, involves comparing outcomes between groups before and after some shock or intervention. Prior to the intervention, those groups must exhibit equivalent trends.

The basic textbook approach compares two groups — one affected (or ‘treated’ in the usual jargon) by an intervention, the other not. The effect of the intervention can thus be estimated by comparing the difference between the treated and untreated group, controlling for any baseline trend exhibited by the untreated group over time. (For example, if there is already a trend increase in a given outcome, then you would want to partial that out when estimating the specific effect of the intervention.)

In the case of France during the Napoleonic Wars, there is no purely untreated group. All of France experiences the blockade in some way. But Juhász exploits differences in the degree of ‘treatment’ — that is, variation across France in the trade cost shock induced by the change in effective distance to Britain.

The principal result is highlighted in column 1 in the table below. In this basic specification, only the effective distance to London is included as an explanatory variable (with time and location fixed effects). There is a strong and significant effect: the greater the effective distance to London, the greater the cotton manufacturing capacity in a given département.

Of course, one might suppose that other factors could explain the change in industrial activity. Juhász thus includes various controls: energy sources (water, coal), proximity to major urban centres (market potential) and human capital (knowledge access, literacy). She tests each control individually (columns 2 through 6) as well as jointly (coumn 7). While the magnitude of the effect of distance changes modestly with different configurations of controls, the overall effect remains significant.

Short-run effects: How change in effective distance influenced mechanised cotton spinning in France, 1803–1812

Dependent variable:
Spindles per thousand inhabitants
Effective distance33.47
Streams × 1812−0.14
Coal × 1812−3.93
Market potential × 181241.05
Knowledge access × 181240.87
Literacy × 181246.41
Time and département fixed effectsYYYYYYY
Adjusted R20.340.330.340.360.370.420.45
Standard errors (clustered at the département level) in parentheses. Effective distance: the natural logarithm of the shortest route between London and each département in, respectively, 1803 and 1812.
Source: Table 1, Juhász 2018.

Additionally, Juhász shows long-run effects of the blockade, finding that the domestic reorganisation of French cotton production persisted through to the late nineteenth century. French cotton spinning capacity remained tightly concentrated in the country’s north several decades after Napoleon’s defeat.

Some loose threads

In the broad, Juhász presents a credible analysis of the effects of the blockade on French cotton manufacturing. There is clear evidence that changes in the cost of acquiring British cotton in France had an effect on French cotton manufacturing. With this trade cost shock being greater in the north of France, it was here that cotton manufacturers experienced the best conditions for domestic expansion.

Nevertheless, there are some smaller points one can quibble over. I am sceptical about Juhász’s measure of knowledge access. The intent of this control variable is to address the possible concentration of individuals with technical know-how relevant for mechanised cotton spinning. This is a key factor that may confound the results: if know-how is already concentrated in the north of France (for example, because of its proximity to Britain), then it would not be surprising to see takeoff in adoption of the technology in the same area.

Juhász’s proxy for each département‘s knowledge access is distance to a university. While I would ordinarily be inclined to accept this as a measure of ‘upper-tail human capital’ — that is, an educated elite with specialised skills — it is not clear to me that universities are the right educational institution in this case. At the start of the nineteenth century, universities were principally institutions cultivating graduates to serve the state or the church; the primary ‘outputs’ were jurists and priests. These are not exactly the skills that would prove useful in adopting new industrial techniques. Other authors have considered advances in French technical education and scientific knowledge, which would be more relevant (though such studies draw on data from later in the nineteenth century).

On a related point, the other human capital measure — literacy — is also somewhat problematic. First, literacy is measured using marriage certificate signatures — the logic being that if you can sign your own name on a form, you can write; if you can write, you can likely read. There are known weaknesses in using marriage certificates as a historical literacy measure (as discussed by Nilsson (1999), among others). But if one assumes that mismeasurement of literacy associated with this proxy is broadly consistent across all observations, then it should still provide a decent basis for capturing the differences in literacy across France.

The literacy measure is, however, less useful in this case because it is observed between 1786 and 1790: some 20 years before the period Juhász studies. The reason for including literacy is to capture the general educational level of the prospective workforce that might be involved in mechanised cotton spinning. If one expects that mechanisation is likely to be driven by relatively younger workers, then the literacy measure here is capturing the wrong generation. And I am not immediately convinced that one can assume that the trend increase in literacy rates over time would be equivalent across France. It may be that literacy rates increased more rapidly in certain areas — say, the north of France — providing a higher-skilled workforce to embrace technological advances.

The above notwithstanding, Juhász offers a rich and interesting study in economic history. It would be too strong to conclude that her article provides definitive proof of the merits of the ‘infant industry’ argument. (Nor does she claim this.) But at a minimum, it shows the validity of the underlying claim that trade barriers can — under certain conditions — have lasting effects on the composition of economic activity.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.