My new paper details the problems of measuring the periphery’s terms of trade in the nineteenth century.
In a previous post I outlined some of the problems encountered by Jeffrey Williamson when he attempted to measure the periphery’s terms of trade in the nineteenth century. I have now uploaded a new ‘Technical Paper’ titled ‘The Periphery’s Terms of Trade in the Nineteenth Century: A Methodological Problem Revisited’, which is a considerably revised version of Chapter 2 of my PhD dissertation. In it I have detailed the methodological issue and why it affects Williamson’s analysis.
I have been trying to get this paper published for some time. First, I tried an early version with the European Review of Economic History, where Williamson first reported his new dataset of the periphery’s terms of trade. It was, however, rejected by the editor following a negative peer review. For the sake of transparency, I have reproduced the review here:
‘In a nutshell, this paper warns that Williamson’s series of terms of trade should not be relied upon. As such, it is not going to be welcome, as most scholars dislike to be told they should not use data. The only positive contribution is a new series of terms of trade for India. Even if this exchange is highly unbalanced (15 indexes scrapped, one only added) the paper would have been worth of consideration if it were not marred by some serious conceptual and factual mistakes
i) the author claims that terms of trade should be measured with domestic prices, inclusive of duties, and quotes Williamson to support this point. The present referee does not feel this bold claim convincing. First, this is not the standard practice, as far as I know. Indexes of imports and exports are routinely computed with unit values from trade statistics. Anyway, there is not an ideal measure. Each measure can be useful, given the analytical purpose. Terms of trade might be useful to analyze the micro analysis of behavior of agents (and thus possibly de-industrialization), while external terms of trade are necessary for the macroeconomic analysis.
ii) the author seems to ignore the basic principle of valuation in trade statistics – i.e. that imports are measured including transport costs to the frontier (CIF) but excluding duties, and a fortiori, transport costs from the frontier to market cities and commercial mark-ups. Thus, contrary to her statement, the indexes computed with unit values from trade statistics (e.g. for Italy and Japan) are bound to differ from her “ideal” index.
iii) the author argues that the main weakness of Williamson’s “worst case scenario” indexes is the omission of transportation costs. However recent works (e.g. Jacks et al 2009) have shown that international freight factors (international transport costs as share of FOB prices of exports) were very low. Thus an adjustment for transport costs would not change qualitatively the results, as the author himself shows about Indonesian series. The results would be radically different only with domestic terms of trade which include domestic transportation costs (usually higher than international ones, if not by water) and commercial mark-ups.
iv) the author ignores another, and possibly bigger, weakness of Williamson’s indexes – the use of a common price index for imports, which is essentially an index of price of exports of manufactures from the United States. This is unlikely to measure accurately prices of imports of any specific country, as their composition differed widely.’