Today, Davis Kedrosky published an interesting substack article well worth reading on whether infant industry protection in late 19th century Canada was a success. The post is well worth reading but I have reservations regarding the conclusion that it was a success is correct. My reluctance does not stem from my knowledge of international trade theory – which I deem elementary for an economist – but rather from my knowledge of Canadian historical data. As such, I want to throw out reasons to be cautious with regards to the actual findings and argue that there is still a crying need to improve and push that literature forward. Simply put, I see the claim of “success” as plausible but not yet convincingly addressed.
Pre-1879 growth is poorly estimated
The ultimate source in Canadian economic history is that of Urquhart (1993). The problem with that source is its price deflator used to convert everything in real output. The component for consumption is lacking a price component for clothing from 1870 to 1900. This would not be problematic if a) clothing was a small share of expenditures; and b) clothing prices behave like other prices. Both assumptions are, however, wrong. Clothing prices fell much faster than other prices – something I documented in Geloso and Hinton (2020) using prices from newspapers across Quebec and Ontario and prices reported by manufacturers in the Canadian Journal of Fabrics. Its only in the late 1880s that clothing prices began changing at rates similar to other prices. Clothing was also more than 20% of consumer expenditures. As such, their exclusion from the implicit price deflator creates a major bias for the 1870s and early 1880s.
As I showed in Geloso and Hinton (2020), incorporating new price evidence shows that real income levels were much lower in the 1870s and 1880s (by as much as 10%) than the Urquhart series. The more important result is that this alters the growth pace as Canada starts from a much lower base in 1870. Indeed, I found that rather than having a middling growth rate amongst rich nations, Canada has the fastest growth of all rich nations between 1870 and 1900.
Why does this matter? Because most of the papers that rely on Urquhart (some but not all) out there relied on the uncorrected figures that show less growth pre-1879 than was truly the case. Proper corrections suggest that 1879 is not as much of a break-point in growth than a statistical artifact. It is also relevant because the best paper out there on nascent industries and protection – Harris, Keay and Lewis in Explorations in Economic History – uses some limited price indexes to estimate real gross output. This is not a fault in their work as they were working with the best there was at the time. However, it is clear that this is a weak point that needs to be discussed as clothing was one of their key industries in terms of leading the results. The differences between the price series they use and mine are noticeable (see figure below). Their was a wholesale price index (with some adjustments) rather than a producer price index (which is closer to what I had in my work with Hinton) to 1885. Moreover, their post-1885 series is based on a very small array of goods (wool hosiery, wool underwear, gingham apron, gingham dress) whose missing values were interpolated from the same wholesale price index pre-1885. How big are the differences? The annual price changes in clothing industry are placed at -1.7% from 1870 to 1879 in Harris et al. (2015) versus -4.6% in Geloso and Hinton (2020). This must have an effect on the results (even though I should point that I really do like the Harris et al. (2015) paper) and which explains why I am still reluctant to jump on the bandwagon.
Output is missing regionally concentrated industries
Because of its monumental size, the Urquhart volume is a reference volume. Few seem to dispute its precision. However, a few years ago I was at a conference of the Canadian Network for Economic History and there was Marvin McInnis – a man not well known outside of Canada but who is well known by all Canadian economic historians and who was deeply involved as a colleague of Urquhart in creating the 1870 to 1926 estimates (notably with respects to agricultural output). At the conference, McInnis gave a keynote address and said that he was still unsatisfied with some issues – notably that maple products and shipbuilding were not well measured (and underestimated). This is not trivial if he is correct as these were two industries that were regionally concentrated in Quebec and Atlantic Canada. Unfortunately, McInnis never did put ink to pen on this and its hard to assess whether this is true for shipbuilding and he did not explicit his concerns in his chapter in the Urquhart volume. However, Urquhart’s book (pp. 425–427) does mention that certain industries such as shipping were not well measured as reports for vessel operators “on the Great Lakes or on the High Seas” were not required. As such, while some larger operators were reported through trade figures, smaller operators were not and fixed assumptions that Urquhart clearly seemed unsatisfied with were made to approximate output.
However, it is worth investigating and withholding judgment for two reasons because of these issues. First, in the historiography, there is a clear agreement that eastern Quebec and Atlantic Canada suffered most from the 1879 tariff. Second, some of these industries (i.e., shipbuilding) are not considered in some key studies such as Harris et al. (2015) or are considered in spite of the problems associated with their constructions (e.g., waterway shipping). This should make one reluctant to embrace the claim that it was a success – there is still some precision steps to be made.
This last criticism is not levelled at the existing literature. It rather applies to what should be the next step in the conversation: the distribution of costs. All estimates of the benefits and costs of the infant industry protection are produced at the national level – no disaggregation at the regional level is produced. Why does this matter? The reason is tied to the well-known and major price differences across regions in Canada (Minns and MacKinnon 2007; Emery and Levitt 2002). The rough pattern is that prices were higher in Atlantic Canada and Quebec than they were in Ontario. As such a nominal loss of a dollar in these provinces represented a much greater loss in real consumption than the same dollar in Ontario (because prices were lower). If protection reshuffled production within Canada – westwards towards Ontario – the loss in wellbeing would be greater than the one measured at the aggregate level. To this day, no one has attempted to produce correctives in the matter. I am not sure if this will matter (and I have a project to do with Casey Pender of Carleton University, but I am swamped with other Canadian economic history papers now), but this is another reason for my reluctance to declare as enthusiastically as Kedrosky does.
Yes, I do find it plausible that the protection may have boosted some domestic industries. It would fit with the Krugman “New Trade Theory” literature. However, I am reluctant to declare a winner here because of these data issues above. I have too frequently seen important results overturned by improved precision (e.g., Kaufmann 2020) to embrace an empirical finding wholeheartedly when the historical data has shortcomings (note: this is why anyone who reads any of my economic history papers will notice that I generally expend considerable time and space on complementary evidence rather than on the empirical evidence alone – I also often take swipes at my own previous work such as in Geloso (2022, fn. 12 for e.g.) against Geloso et al. 2017).
Emery, J. H., & Levitt, C. (2002). Cost of living, real wages and real incomes in thirteen Canadian cities, 1900–1950. Canadian Journal of Economics/Revue canadienne d’économique, 35(1), 115-137.
Geloso, V., & Hinton, M. (2020). Improving Deflators for Estimating Canadian Economic Growth, 1870-1900. Research in Economic History, 36, 125-150.
Geloso, V., Hinton, M., & Kufenko, V. (2017). The equally “bad” French and English farmers of Quebec: New TFP measures from the 1831 census. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 50(3), 170-189.
Geloso, V. (2022). Unenlightened peasants? Farming techniques among French-Canadians, circa 1851. Cliometrica, 1-23.
Harris, R., Keay, I., & Lewis, F. (2015). Protecting infant industries: Canadian manufacturing and the national policy, 1870–1913. Explorations in Economic History, 56, 15-31.
Kaufmann, D. (2020). Is deflation costly after all? The perils of erroneous historical classifications. Journal of Applied Econometrics, 35(5), 614-628.
Minns, C., & MacKinnon, M. (2007). The costs of doing hard time: a penitentiary‐based regional price index for Canada, 1883–1923. Canadian Journal of Economics/Revue canadienne d’économique, 40(2), 528-560.
Urquhart, M. (1993). Gross national product, Canada, 1870-1926: The derivation of the estimates. McGill-Queen’s Press-MQUP.