making historical international trade data useful · making historical international trade data...
TRANSCRIPT
Making historical international trade data usefulGerman product-level data from 1880 to 1913 and the
Standard International Trade Classification
Wolf-Fabian Hungerland and Chris Altmeppen
Humboldt-Universitat zu Berlin
September 2016
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 1 / 46
Introduction
In a nutshell
B Problem: Have 41 volumes of product level trade data from thelate nineteenth and early twentieth century, almost all usingdifferent product classifications.
B Solution: Make data useful by standardising it into SITC.B We argue that translating historical trade data into SITC is generally
possible and that the translation yields useful results, even on avery disaggregated level. But a lot of manual work is necessary.
B We present guidelines when judgement is necessary.B We then evaluate our translation exercise quantitatively (work in
progress).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 2 / 46
Data Project
Our project
B Digitalise, harmonise and analyse German product-level exportsand imports from 1880 to 1913 (probably interwar too).
B From 1880 to 1888, statistics from Bremen and Hamburg (out ofZollverein) have to be integrated as well.
B To be able to so: Classify products according to the StandardInternational Trade Classification (SITC), ideally at the five-digit, i.e.product-level where possible.
B Collect tariffs at product-level (work in progress, Geschonke 2016).B We have „60,000 products « 41 volumes ˆ „1450 product
categories on average per volume
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 3 / 46
Data Project
Source example: Exports to British South Africa
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 4 / 46
Data Project
Source example: Exports to British South Africa
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 4 / 46
Data Project
Source example: Exports to British South Africa
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 4 / 46
Data Project
Source example: Exports to British South Africa
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 4 / 46
Data Project
Handling a year ˆ product ˆ country panel, by volume
country dimension
tim
ed
ime
nsio
n
Volume Y
Volume X
Year 1
Year 2
Year 3
Year 4
Year 5T
rade
part
ner
1
Tra
de
part
ner
7
Tra
de
part
ner
2
Tra
de
part
ner
3
Tra
de
part
ner
4
Tra
de
part
ner
5
Tra
de
part
ner
6
Product 4
Product 3
Product 2
Product 1
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 5 / 46
Data Project
Problems to solve
B Product categories (’Warengattungen’) between different volumesof the German trade statistics are among themselves difficult if notimpossible to compare without some manipulation.
B Nearly every volume of the German statistics has a differentclassification of products.
B The number of categories ranges from roughly 500 in the 1880s tomore than 2300 in the 1910s.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 6 / 46
The SITC A primer on SITC
SITC: 1-digit levels
Section Description0 Food and live animals1 Beverages and tobacco2 Crude materials, inedible, except fuels3 Mineral fuels, lubricants and related materials4 Animal and vegetable oils, fats and waxes5 Chemicals and related products, n.e.s.6 Manufactured goods classified chiefly by material7 Machinery and transport equipment8 Miscellaneous manufactured articles9 Commodities and transactions not classified
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 7 / 46
The SITC A primer on SITC
SITC: An example
Code DescriptionSection 0 Food and live animalsDivision 05 Vegetables and fruitGroup 054 Vegetables, fresh, chilled, frozen or sim-
ply preserved (including dried leguminousvegetables); roots, tubers and other ediblevegetable products, n.e.s., fresh or dried
Subgroup 0545 Other fresh or chilled vegetablesBasic Heading 05458 Mushrooms and truffles, fresh or chilled
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 8 / 46
The SITC A primer on SITC
The SITC is helpful, ...
B Allows consistent aggregation.B Allows standardised comparison over time and space.B Linkable to other classifications, let it be from trade (HS, BEC) or
other (ISIC, Lall, Rauch).B Is the analytical standard.B If revisions are pooled, almost 6000 categories are available.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 9 / 46
The SITC A primer on SITC
... but sometimes hard to handle
B Which revision of the four to use?B How to concord?B How to maintain details?B Sometimes even: How to get the right tables?B Scientific literature?
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 10 / 46
Applying the SITC to our data Framework
We classified (almost) all products we found into SITC
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 11 / 46
Applying the SITC to our data Framework
We are already far in ”SITCising” our data
0 500 1,000 1,500 2,000 2,500Available product categories in volume
Reich
Hamburg
Bremen
KStA (1914, 271)KStA (1912, 252)
KStA (1907, 182 [Mar.-Dec.])KStA (1907, 182 [Jan.-Feb.])
KStA (1906, 172)KStA (1903, 152)
KStA (1897, 91)KStA (1896, 85)KStA (1895, 79)KStA (1894, 73)KStA (1893, 67)KStA (1892, 61)KStA (1891, 55)KStA (1890, 48)KStA (1889, 41)KStA (1888, 34)KStA (1887, 26)KStA (1886, 20)KStA (1885, 15)KStA (1884, 10)
KStA (1883, 61.2)KStA (1882, 55.2)KStA (1881, 50.2)
HStB (1888)HStB (1887)HStB (1886)HStB (1885)HStB (1884)HStB (1883)HStB (1882)HStB (1881)HStB (1880)
JfBS (1888)JfBS (1887)JfBS (1886)JfBS (1885)JfBS (1884)JfBS (1883)JfBS (1882)JfBS (1881)JfBS (1880)
KStA (1903, 152), KStA (1906, 172), KStA (1912, 252) and KStA (1914, 271) cover multiple years.
SITCised not (yet) SITCised
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 12 / 46
Applying the SITC to our data Practical problems
Problems: Overview
1. Vagueness2. Missing information3. Different grouping of goods4. Unclear stages of production5. Usage vs. consumption6. Production methods7. Revision mess
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 13 / 46
Applying the SITC to our data Practical problems
Problem 1: Vagueness
B Problem: Many goods cannot be assigned to 5-digit SITC codesbecause they combine goods of several categories or are toovague.
B Solution: If a reasonable assumption about a dominant good inthe category can be made, assign the best fitting 5-digit SITCcategory. Otherwise try a higher level SITC category.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 14 / 46
Applying the SITC to our data Practical problems
Problem 1: Vagueness – example 1
B The German product nr. Geraeth (equipment, devices) is veryvague and impossible to assign on a detailed SITC category.
B We assign the very broad SITC section 7 (Machinery and transportequipment).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 15 / 46
Applying the SITC to our data Practical problems
Problem 1: Vagueness – example 2
B Wall- und Haselnusse (Wal- and Hazelnuts) are grouping SITC05776 (Walnuts) and 05775 (Hazelnuts).
B We assign the 4-digit category 0577 (edible nuts).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 16 / 46
Applying the SITC to our data Practical problems
Problem 1: Vagueness – example 3
B Feigen, getrocknete (dried figs) can be assigned to SITC 0576 (figs,fresh or dried).
B However, there is sometimes another category for fresh figs andpistachio which have different SITC categories. Since pistachioshave the code 05778 in SITC, we could assign the aggregatedcategory 057 (fruits and nuts, fresh or dried).
B But since we see the pistachios category only in some volumeswhile figs occur in most volumes, we assume that figs are thedominant good in this category and assign them to SITC 0576.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 17 / 46
Applying the SITC to our data Practical problems
Problem 2: Missing information
B Problem: Sometimes the information required to assign the correctsubcategory is just missing.
B Solution: Try to make a reasonable assumption or to getinformation on products in contemporary encyclopaedias orhandbooks.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 18 / 46
Applying the SITC to our data Practical problems
Problem 2: Missing information – example 1
B The 4-digit SITC code 0011 (Bovine animals, live) is further dividedinto 00111 (Pure-bred breeding animals) and 00119 (Other thanpure-bred breeding animals).
B However, in the older volumes of the StdDR, it is mostly impossibleto determine if the specified bovine animals are for breeding or not.Kuhe (Cows) can technically only be assigned to 0011, whileOchsen (ox) can technically be assigned to 00119, since ox arecastrated cows.
B We assume that the vast majority of traded cattle is not for breedingand therefore assign 00119 to all of them.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 19 / 46
Applying the SITC to our data Practical problems
Problem 2: Missing information – example 2
B In SITC 0111 (Meat of bovine animals, fresh or chilled) is furtherdivided in whether it contains bones or not.
B In the StdDR this information is not given.B Since we cannot make a reasonable assumption on the
containment of bones, we assign 0111 to all categories.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 20 / 46
Applying the SITC to our data Practical problems
Problem 2: Missing information – example 3
B The SITC 248 (Wood, simply worked, and railway sleepers ofwood) categories differentiate between coniferous wood (24823)and other wood (24845).
B Many German product categories do not provide detailedinformation about the kind of woods.
B We therefore assigned them the 3-digit level.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 21 / 46
Applying the SITC to our data Practical problems
Problem 3: Different grouping of goods
B Problem: Different approaches in dividing products into subgroups,e.g. by characteristics of their material, processing or usageimpede precise matching.
B Solution: Make reasonable assumptions or use higher aggregatedSITC-levels.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 22 / 46
Applying the SITC to our data Practical problems
Problem 3: Different grouping of goods – example 1
B The SITC distinguishes between clothes that are not knitted orcrocheted, for men (841) and women (842), and clothes that areknitted or crocheted, for men (843) and women (844).
B The StdDR does distinguish between men and women’s clothes inmost cases, but not always between the knitted or non-knittedclothes. In the case of Leibwasche (underwear), there is noinformation given for whether the fabrics is knitted, nor whether it ismade for men or women.
B We assign it the two-digit level.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 23 / 46
Applying the SITC to our data Practical problems
Problem 3: Different grouping of goods – example 2
B Sometimes products in the German trade data are separated bycountry of origin, which is usually motivated by tariff discriminationover different countries. Weizen, russischen Ursprungs (wheat,Russian origin) and Weizen, nicht russischen Ursprungs (wheat,non-Russian origin) are an example of that.
B We assigned SITC 041 (Wheat (including spelt) and meslin,unmilled) and include a dummy for a hypothesized protecionistagenda.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 24 / 46
Applying the SITC to our data Practical problems
Problem 3: Different grouping of goods – example 3
B Some products in the German trade data are differentiated to adegree that cannot be reproduced by the SITC categories.
B As usual in that time, when cotton played a much more central rolein international trade, yarn products often declare the degree offineness of the yarn or whether it is single or multiple twisted.
B Since the SITC does only state the material, we have to assign thesame category to all respective German product numbers:
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 25 / 46
Applying the SITC to our data Practical problems
Problem 4: Unclear stages of production
B Problem: SITC is to some degree structured according to thestages of production.
§ Crude materials are described in section 2.§ Manufactured goods can be found in section 6, where they are
classified by material and mostly simply manufactured.§ Section 8 classifies mostly by area of utilization and usually are more
complex.§ Sometimes is not clear on which stage of production the products
should be classified.B Solution: If we are not able to make an educated guess about the
stage of production, we assume the latest stage of productionpossible.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 26 / 46
Applying the SITC to our data Practical problems
Problem 4: Unclear stages of production – example 1
B Seidenwaaren (silk goods) could in terms of the SITC include rawsilk (not thrown) (2613), silk yarn (65192) or fabrics woven of silk(65411).
B We assigned the latter category:
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 27 / 46
Applying the SITC to our data Practical problems
Problem 4: Unclear stages of production – example 2
B Pelzwerk (fur, articles thereof) has corresponding categories inSITC sections 21 (Hides, skins and furskins, raw), 61 (Leather,leather manufactures, n.e.s., and dressed furskins) or 84 (Articlesof apparel and clothing accessories).
B We assigned the SITC category 84831 (Articles of furskin).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 28 / 46
Applying the SITC to our data Practical problems
Problem 4: Unclear stages of production – example 3
B Holzwaaren (articles of wood) could include very raw as well asvery sophisticated products. Since other product names startingwith Holzwaaren in this volume are clearly not raw materials, e.g.Holzwaaren, Kisten u. Koerbe (articles of wood, boxes andbaskets).
B Wwe assign the SITC category 635 (Wood manufactures, n.e.s.).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 29 / 46
Applying the SITC to our data Practical problems
Problem 5: Usage vs. consumption
B Problem: The usage of a product can determine its appropriateSITC category.
B Solution: Do research about historical usages and consumptionhabits to find the best matching category.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 30 / 46
Applying the SITC to our data Practical problems
Problem 5: Usage vs. consumption – example 1
B The rise of fertilizers in the 19th century is reflected in the StdDRproduct categories.
§ Thomasschlacke (Thomas-slag) is a slag from the manufacture ofiron and steel. It was used as a fertilizer in its granulated form.Nowadays, the metallurgic process (Thomasverfahren) and itsbyproduct are no longer used.
§ The relevant SITC categories could therefore be 2786 (slags, dross,scalings and similar waste) or something in division 56 (Fertilizers).
B Because we are primarily interested in the historical productionstructure, we assign SITC code 27862 (granulated slag from themanufacture of iron or steel). An assignment of a fertilizer categorywould suggest a different production structure when the data arelater used for analysis.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 31 / 46
Applying the SITC to our data Practical problems
Problem 5: Usage vs. consumption – example 2
B Some products in the data that were used as foodstuffs would notbe regarded as such today. Bleizucker (lead sugar), for example,was produced synthetically and used as a sweetener already inancient times, but became illegal in the 19th century due to itstoxicity.
B We assign Bleizucker to SITC 51371 (Acetic acid and its salts)since we assume that it was mostly not used as a sweetener andwe do not lose accuracy and information about production with thiscategory.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 32 / 46
Applying the SITC to our data Practical problems
Problem 5: Usage vs. consumption – example 3
B The same is true for the Spanische Fliege (Spanish fly), which wasused as an aphrodisiac, but is not common today for the samereasons as Bleizucker.
B Spanische Fliege was assigned to category 29198 (Ambergris,castoreum, civet and musk; cantharides; bile, whether or not dried;glands and other animal products used in the preparation ofpharmaceutical products, fresh, chilled, frozen or otherwiseprovisionally preserved), since other categories such as 29199(Animal products, n.e.s.) would lead to a loss of accuracy andinformation.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 33 / 46
Applying the SITC to our data Practical problems
Problem 5: Usage and consumption – example 4
B Brechweinstein (emetic tartar) was a common emetic during the19th century, but due to its toxicity its only used as a mordant in thetextile industry.
B We still assign SITC code 54293 (Medicaments, n.e.s., put up inmeasured doses or in forms or packings for retail sale) to it, sincethe fluctuation in usage of chemicals as medicaments is inevitable.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 34 / 46
Applying the SITC to our data Practical problems
Problem 6: Production methods
B Problem: During the 19th century production processes changed alot for many products, especially with the rise of the chemicalindustry in Germany during that time. This is relevant for theclassification, but we often do not know how the products describedwere produced.
B Solution: Find information about the stance of contemporaneoustechnology.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 35 / 46
Applying the SITC to our data Practical problems
Problem 6: Production methods – example 1
B Another prominent and heavily traded fertilizer during that time wasguano, the excrement of seabirds.
B Guano: kunstlicher (artificial Guano) suggests a category indivision 56.
B But neither the method of production, nor the composition is givenin the data. On the world exhibition of 1876, artificial guano wasadvertised by the Pacific Guano Company (CITATION). It wasproduced by using lime from South Carolina which was combinedwith fish cadavers.
B This clearly leads to assigning SITC 2721 (Animal or vegetablefertilizers, whether or not mixed together or chemically treated;fertilizers produced by the mixing or chemical treatment of animalor vegetable products).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 36 / 46
Applying the SITC to our data Practical problems
Problem 6: Production methods – example 2
B Production processes changed over time, especially with the rise ofthe chemical industry in Germany during that time.
B Campher for example was extracted from plants before itssynthesis was discovered in 1903 and industrial production beganin 1907. There is no information in the data whether the tradedcampher was synthetically or naturally produced.
B In SITC there is only a specific category for campher in the division51 (organic chemicals). Naturally produced campher wouldtechnically fall under division 29 (Crude animal and vegetablematerials, n.e.s.), but since the category 51627 (camphor)preserves the information about the product better, we chose that.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 37 / 46
Applying the SITC to our data Practical problems
Problem 6: Production methods – example 3
B Leim is a certain type of glue (paste). It can be made syntheticallynowadays but was produced from animal materials for most time inhistory.
B Therefore we assigned SITC code 59224 (Gelatin (including gelatinin rectangular, including square, sheets, whether or notsurface-worked or coloured) and gelatin derivatives; isinglass; otherglues of animal origin, n.e.s.).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 38 / 46
Applying the SITC to our data Practical problems
Doubtful classifications happen, but not as often asone thinks
67.13%
9.842%
23.03%
Clearly classified Doubtfully classified Not (yet) SITCized
B Documentation of rules and doubts necessary!
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 39 / 46
Applying the SITC to our data Practical problems
Problem 7: Revision mess
B Which revision to use?B We assume: Different SITC numbers may cover the same thing,
but no number cover different things in different revisions.B Goal: Always get most precise, i.e. 5-digit number
1034
1468
28222652
Σ 1912
Σ 2575
Σ 4195Σ 4016
0
700
1400
2100
2800
3500
4200
Rev. 1 Rev. 2 Rev. 3 Rev. 4
5-digit 4-digit 3-digit 2-digit 1-digit
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 40 / 46
Applying the SITC to our data Practical problems
Problem 7: Revision mess – We used mostly revision4
4.326%1.688%
92.79%
rev_total_1 rev_total_2rev_total_3 rev_total_4
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 41 / 46
Applying the SITC to our data Practical problems
Problem 7: Revision mess – We used mostly revision4
0 500 1,000 1,500 2,000 2,500Available product categories in volume
Reich
Hamburg
Bremen
KStA (1914, 271)KStA (1912, 252)
KStA (1907, 182 [Mar.-Dec.])KStA (1907, 182 [Jan.-Feb.])
KStA (1906, 172)KStA (1903, 152)
KStA (1897, 91)KStA (1896, 85)KStA (1895, 79)KStA (1894, 73)KStA (1893, 67)KStA (1892, 61)KStA (1891, 55)KStA (1890, 48)KStA (1889, 41)KStA (1888, 34)KStA (1887, 26)KStA (1886, 20)KStA (1885, 15)KStA (1884, 10)
KStA (1883, 61.2)KStA (1882, 55.2)KStA (1881, 50.2)
HStB (1888)HStB (1887)HStB (1886)HStB (1885)HStB (1884)HStB (1883)HStB (1882)HStB (1881)HStB (1880)
JfBS (1888)JfBS (1887)JfBS (1886)JfBS (1885)JfBS (1884)JfBS (1883)JfBS (1882)JfBS (1881)JfBS (1880)
KStA (1903, 152), KStA (1906, 172), KStA (1912, 252) and KStA (1914, 271) cover multiple years.
Rev. 1 Rev. 2 Rev. 3 Rev. 4
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 41 / 46
Applying the SITC to our data Practical problems
Problem 7: Revision mess – concordance
B We are working on a decent concordance table and an SITC ”shelf”
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 42 / 46
Applying the SITC to our data Practical problems
Problem 7: Revision mess – our SITC ”shelf”
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 43 / 46
Applying the SITC to our data Results
Precision
0 1 2 3 4 5Average length of SITC number (variable: k_max)
Reich
Hamburg
Bremen
KStA (1914, 271)KStA (1912, 252)
KStA (1907, 182 [Mar.-Dec.])KStA (1907, 182 [Jan.-Feb.])
KStA (1906, 172)KStA (1903, 152)
KStA (1897, 91)KStA (1896, 85)KStA (1895, 79)KStA (1894, 73)KStA (1893, 67)KStA (1892, 61)KStA (1891, 55)KStA (1890, 48)KStA (1889, 41)KStA (1888, 34)KStA (1887, 26)KStA (1886, 20)KStA (1885, 15)KStA (1884, 10)
KStA (1883, 61.2)KStA (1882, 55.2)KStA (1881, 50.2)
HStB (1888)HStB (1887)HStB (1886)HStB (1885)HStB (1884)HStB (1883)HStB (1882)HStB (1881)HStB (1880)
JfBS (1888)JfBS (1887)JfBS (1886)JfBS (1885)JfBS (1884)JfBS (1883)JfBS (1882)JfBS (1881)JfBS (1880)
KStA (1903, 152), KStA (1906, 172), KStA (1912, 252) and KStA (1914, 271) cover multiple years.
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 44 / 46
Applying the SITC to our data Results
Sectoral coverage (1-digit level)
0 500 1,000 1,500 2,000 2,500Available SITC product categories in volume
Reich
Hamburg
Bremen
KStA (1914, 271)KStA (1912, 252)
KStA (1907, 182 [Mar.-Dec.])KStA (1907, 182 [Jan.-Feb.])
KStA (1906, 172)KStA (1903, 152)
KStA (1897, 91)KStA (1896, 85)KStA (1895, 79)KStA (1894, 73)KStA (1893, 67)KStA (1892, 61)KStA (1891, 55)KStA (1890, 48)KStA (1889, 41)KStA (1888, 34)KStA (1887, 26)KStA (1886, 20)KStA (1885, 15)KStA (1884, 10)
KStA (1883, 61.2)KStA (1882, 55.2)KStA (1881, 50.2)
HStB (1888)HStB (1887)HStB (1886)HStB (1885)HStB (1884)HStB (1883)HStB (1882)HStB (1881)HStB (1880)JfBS (1888)JfBS (1887)JfBS (1886)JfBS (1885)JfBS (1884)JfBS (1883)JfBS (1882)JfBS (1881)JfBS (1880)
KStA (1903, 152), KStA (1906, 172), KStA (1912, 252) and KStA (1914, 271) cover multiple years.
Food & animals Beverages & tob. Crude materials Mineral fuelsAnimal & veget. oils Chemicals Manufact. goods MachineryMisc. manufact. NES not (yet) SITCised
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 45 / 46
Summary
Summing up
B First and foremost: This is still work in progress.B Using SITC for historical data looks very promising (judging by
precision and coverage), ...B however transparency as we well as proper documentation of rules,
doubts and hypotheses is, of course, necessary.B Expect to invest a lot of work .B Revision 3 seems richest (though we used mostly revision 4 and
work on a concordance).
Hungerland & Altmeppen (HU Berlin) Historical data and SITC September 2016 46 / 46