English/British/UK Trade Statistics, 1700-1899 David Jacks (SFU and NBER) September 2016
Roadmap (I) Provide an overview of English/British/UK trade statistics from 1700 to 1899 (emphasis on qualitative aspects of data and collection, not on quantitative analysis) (II) Provide an example of how we intend to use these trade statistics with a particular eye towards classification 2
(I) Overview of EN/GB/UK trade statistics Primary source: Customs and Excise Department’s ledgers of imports and exports, published under various titles from 1697 to 1899. Same data underlying Schlote (1938) and Schumpeter (1960), but not aggregated. Closest analog: Davis (1979) with 15 partners for 20 commodity groups from 1785 to 1855. 3
First objective of original data collection was in the determination of flows for revenue purposes. Naturally by 1697, there was already an extensive set of excises and tariffs in place. Also a wide appreciation of role of external markets in shaping course of domestic economy. Thus, a secondary objective in using balance of trade as a vital sign for English economy. 4
The ledgers contain a line-by-line account of the commodity-level bilateral trade flows with the rest of the world; universe of trade(?). The data is further disaggregated for some periods by distinguishing between trade with London versus the “ outports ” and trade carried out on domestic versus foreign ships. Units: England up to 1780, Great Britain up to 1830, the United Kingdom up to 1899. 5
6
Our goal: fully replicate structure of the ledgers by recording (and analyzing) the commodity-level trade with all trading partners. Sampling at every decade from 1700 to 1899; thus, our sample encompasses: (1) last decades of mercantilist era (1700-1780); (2) the IR, the FWs, and unilateral move towards free trade (1780-1850); (3) first wave of globalization and non-British industrialization (1850-1899). 7
However, the data do not come without certain caveats: (1) Before 1904, exports were reported for those countries to which they were directly shipped while imports were reported as such for those countries from which there were directly shipped. Thus, transshipment trade is unreported and land-locked nations are absent in the data. 8
(2) Some have disparaged the use of the ledgers as accurate guides to the commercial life of Britain (Clark 1938). This criticism relates to question of reliability of effort and of procedure (more below) but follows a large historical literature on prevalence of smuggling and under-reporting of imports. This latter concern while valid is likely a small one… 9
The scale of this type of activity was almost certainly dominated by the volume of trade legitimately recorded. More importantly, nothing in the historical record suggests that any related bias is systematic across nations. However, commodity composition might matter in this regard and more work needs to be done. 10
(3) Underlying prices used to value exports were fixed from 1702 to 1813 and from 1702 to 1853 in the case of imports. These seem to be based on declared prices prevailing in or around 1700. Prices were determined by the Customs & Excise Department and were an attempt to minimize bargaining and maximize compliance(?). 11
On the one hand, ledgers constitute a highly consistent record of physical quantities imported and exported (e.g., Deane and Cole, 1962). In this sense, the English/British/UK data is uniquely complete relative to it many of it counterparts. Likely reflects administrative capacity of the state at this time and Britain’s unique geographic circumstances of being an island nation. 12
On the other, Davis (1979) suggests that combining quantities with price information could lead to reliable measures of value of trade. At the time, Davis could only rely upon the price data underlying Gayer et al. (1953), which limited his focus to the years from 1785. There have been a few advances in intervening years, especially collection of previously published and newly discovered price data. 13
One obvious starting point is various collections online, e.g. GPIH (UCD), SHE (Utrecht). Another innovation in this respect is Clark (2005) which documents trends in prices and wages for the English economy from 1209 to 2004. We had hoped to match price/trade data at commodity level, but gaps too numerous… need for aggregation (and/or classification). 14
(4) What is more, large changes come not from the data collection per se but from the evolving structure of the British economy. Some goods categories emerge and some disappear while new trading partners are also registered in the data. The chief obstacle in this respect is not in assessing reliability of data but in assigning consistent good/partner categories across years. 15
16
17
18
We first apply modern country definitions: from the 134,918 raw observations on exports we arrive at 101,808 observations with consistent country identifiers (doing same for M and ReX). We then assign SITC codes to these: (1) One digit for all obs. (2) Two digits for 97950 obs. (3) Three digits for 5210 obs. (4) Four digits for 5037 obs. (5) Five digits for 173 obs. 19
(II) Example of use with categorization Primary question: How well do standard variables drawn from contemporary literature explain bilateral exports during this period? In particular, does distance matter or are trade flows determined by the presence of empire? And do these effects change over time? 20
Bilateral trade flows characterized by: G M M ex im x . ijt t it jt ijt That is, they are a function of factors common to all countries, factors within particular countries such as size and productivity, and factors specific to country-pairs. Country-pair specific factors of particular interest here as they are thought to capture the bilateral trade costs facing countries. 21
Specification is, of course, problematic if we only observe (aggregate) bilateral trade as an exact correspondence with time-varying exporter or importer fixed effects emerges. But we can run this using commodity-level data and retain time-varying fixed effects. However, there is the slight problem that our observations on exports have fixed prices until 1820 and market (?) prices from 1820. 22
The way we work around this is by including SITC one- and two-digit SITC codes interacted with time to capture price effects. Provided that products in one- and two-digit categories to a first approximation vary together, these fixed effects cast export flows in real terms (prices not allowed to change, but quantities can). So, we now estimate: im x G M M (exp(distance empire ). ijt t jt kt jt jt 23
24
Recommend
More recommend
