The devil is in the details: assessing early industrial performance across international borders using late nineteenth century North American manufacturers as a case study

Abstract

Was the average late nineteenth century Canadian manufacturer a hopelessly inefficient laggard when compared to his American counterparts just on the other side of the international border? Could it have been that Canadian producers were really productivity leaders so early in their industrialization process? To answer these fundamental questions about the performance of a small open economy at a key juncture in its industrial development we need a quantitative comparison of Canadian relative to US productivity. Unfortunately, the approach we adopt with respect to data compilation, methodological choice, and the identification of inputs, outputs, input weights, and prices has both quantitative and qualitative effects on any performance assessment we may undertake. To illustrate the severity of these effects we use manuscript census data from 1870/1871 to derive a series of total factor productivity (TFP) comparisons for a sample of manufacturing establishments located immediately to the north and south of the Canada-US border. Even with some of the most common and defensible variable definitions, we find dramatic differences between the most optimistic and most pessimistic assessments of late nineteenth century Canadian performance. Our case study suggests to us that those interested in the assessment of long run performance differentials should appreciate the sensitivity of these assessments and they should carefully consider the approach that has been taken toward data preparation, methodological choice, and variable definitions in their construction.JEL codes D2 Á L6 Á N6

Firm heterogeneity, technology adoption and the spatial distribution of population: Theory and measurement

“…However, this comparison is somewhat misleading because differences in the enumeration practices between the Canadian and US censuses overstate the magnitude of this difference. The US census did not enumerate establishments with revenue of less than $500, whereas Canadian enumerators were instructed to enumerate all establishments regardless of size (Inwood and Keay 2008). When the Canadian sample is censored to exclude establishments falling below the US census revenue threshold, the fraction of establishments using steam power increases to 15.7%.…”