DATA
The estimates are in CSV format and can be downloaded HERE. The description of variables is provided in this README.txt.The information covers the years 2012-2019.
You can also access the data directly by executing the following commands in your preferred language.
Home
Research
Books & Notes
Coding Guides
Data
T.E.C.H. (Blog)
using CSV, Downloads, DataFrames
dataset = DataFrame(CSV.File(Downloads.download("https://alfaromartino.github.io/data/RevenueManufacture_NACE4.csv")))dataset <- read.csv("https://alfaromartino.github.io/data/RevenueManufacture_NACE4.csv")import pandas as pd
dataset = pd.read_csv("https://alfaromartino.github.io/data/RevenueManufacture_NACE4.csv")
Eurostat's dataset sbs_na_ind_r2 provides official statistics for revenue in the manufacturing sector (sector C, with codes from 1000 to 3399 as defined by the NACE rev. 2 classification). They come from its Structural Business Statistics (SBS). The issue is that these values aren't reported for all industries, due to confidentiality matters.
Attending to this, I provide revenue estimates for several European countries at the NACE (rev. 2) 4-digits level. The data are at the cross-section level, for each year between 2012-2019. The completion for each year is based on an iterative procedure, using Eurostat's information from other years and the ORBIS dataset.
Based on Eurostat's revenues for that specific year, I identify:
each sector's revenue share relative to manufacturing,
each industry's revenue share relative to its sector.
Taking those shares, I compute the missing shares in (a) and (b). The completion of shares is performed separately by industries and sectors to improve accuracy. This means that any imprecision at the industry level is not transmitted to the sector shares. Specifically, I compute missing shares by:
using relative shares from contiguous years in Eurostat's data,
if there are still missing shares, I use information from ORBIS. This exploits that ORBIS reports each firm's revenue at the NACE 4-digits level, making it possible to compute any remaining relative share at the industry and sector level.
The estimates are especially relevant for the years 2016-2018, since ORBIS' information has been improving over the years (2019 has the issue of only using previous years from Eurostat rather than contiguous ones, to avoid using data during the pandemic). Furthermore, the richness of ORBIS depends on the country considered, since its coverage varies by country.
ORBIS coverage is particularly rich for the following countries: Bulgaria, Croatia, Czech Republic, Finland, France, UK, Hungary, Italy, Norway, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia, Spain, and Sweden. Despite its importance, Germany's coverage in ORBIS is relatively low. The next graph shows each country's revenue coverage in ORBIS relative to Eurostat (year 2018).
