DANTRADE
Danish trade and tariff statistics, 1830–2000
| Principal investigators |
|
| Funder |
|
| Period |
|
| Partner |
|
| Collaborators |
|
| BDAD role | Christian Møller Dahl — the automated transcription component |
For 170 years the Danish state published, annually and in detail, what the country bought and sold: commodity by commodity, partner country by partner country, with the tariff rates applied. The volumes survive in full. They are also printed tables running to hundreds of thousands of rows, which means that in practice economic historians have used scattered samples of them and little more.
DANTRADE digitises the series in its entirety and publishes it as structured, machine-readable data — a continuous account of Danish trade and trade policy from 1830 to 2000, covering the country's transformation from an agrarian economy into an industrial one, two world wars, the interwar collapse of world trade, Bretton Woods, and accession to the European Communities.
Why the whole series matters
The questions that need this data are precisely the ones a sample cannot answer. How trade policy responded to shocks and how quickly; which sectors were protected, when, and with what consequence for their productivity; how a small open economy adjusted to the successive collapses and reconstructions of the international trading order. Answering those requires the years between the famous ones, at full commodity detail, which is exactly what has been missing.
BDAD's contribution
The unit contributes the automated transcription layer. Printed historical tables are easier than handwriting but not easy: the typography changes across decades, columns are reorganised, commodity classifications are revised and renumbered, and a single misaligned column silently corrupts a whole year. The methods used here — layout analysis, table-structure recognition, and cross-model agreement used as a quality signal — come directly from the unit's tabular document digitisation work.
Statistics Denmark scans the volumes and hosts the finished data, which means the output is available on the national statistical agency's own terms rather than as a project download that decays when the grant ends.
| A note on credit. DANTRADE is led by Paul Sharp, not by BDAD; Christian Møller Dahl contributed the automated transcription component. It appears here because the unit's methods are a substantive part of the infrastructure being built, and because it illustrates how BDAD normally works — as the data-construction component of a colleague's research programme. It is also the closest collaboration on this page, since HEDG and BDAD share a department and a corridor. |