An estimate of Argentine industrial output from the 1870s to 1913 illustrates a problem with the New Economic History.

The ‘New Economic History’ has sought to transform the study of history by applying econometric techniques to the past. As such, it has greatly increased the demand for historical statistics. The problem has been the supply, as there simply are not enough good quality data to apply econometric techniques to historical issues.

In a previous post, I described D.C.M. Platt’s warning about his colleagues increasing use of ‘Mickey Mouse numbers’. In a new working paper titled ‘(Mis)measuring Argentina’s Progress: Industrial Output, 1870s-1913’ I have now provided a detailed case study from my research on Argentina. The working paper shows that Roberto Cortés Conde’s estimates of Argentina’s industrial output during this period are fundamentally flawed, yet they have been highly influential because they are the only estimates available.

The working paper is based on an attempt to replicate Cortés Conde’s results – a technique that is commonly used in science, but more rarely in the social sciences and humanities. I found that Cortés Conde apparently mistook an increase in the quantity of goods being taxed for an increase in the quantity being produced. This undermines his results for the 1890s, in particular, when the internal taxes on industrial production were introduced. As the scope of the internal taxes widened and their collection became more efficient, a greater quantity of goods were taxed. Cortés Conde’s mistake was to treat this as equivalent to an increase in production. For wine, in particular, this is a major error, as during 1892-97 only fortified wines were taxed, but in 1898 the levy was extended to so-called ‘natural wines’, which made up the vast bulk of production. From looking at Cortés Conde’s methodological notes, original sources, and what he has revealed about his results, this was the only way I could explain the high growth rates that he found. I also looked at a range of alternative indicators that suggest a lower growth rate, although I argue in the paper that even these should only be considered as indicative, given the lack of data.

From this Argentine case study, I extrapolate by arguing that one of the problems with the New Economic History is that it has increased the demand for historical statistics despite the intrinsic limitations in their supply, thereby making such Mickey Mouse numbers appear far more valuable than they actually are.