Workstream (D) – Impact Analysis - European Commission · 1ESSnet on Linking of Microdata on ICT Usage, Grant number 50701.2010.001-2010.578 and ICT Impacts Grant number 49102.2005.017-2006.128

INTERIM REPORT

Including

A Note on Deflators ESSLait Metadata Repository

ESSnet on Linking of Microdata to Analyse ICT Impact

Eurostat Grant Agreement 50721.2013.001-2013.082

May 2013

Authors Eva Hagsten

Michael Polder Ekaterina Denisova

Eric Bartelsman Patricia Kotnik

For further information please contact: Eva Hagsten Michael Polder

Statistics Sweden Statistics Netherlands [email protected] [email protected]

mailto:[email protected]


http://www.essnet-portal.eu/

http://www.essnet-portal.eu/

Interim Technical Implementation Report ESSnet on Linking of Microdata to Analyse ICT Impact


2

Introduction Eva Hagsten, Statistics Sweden (SCB)

This is the interim technical implementation report of the ESSnet on Linking of Microdata to

analyse ICT Impact (ESSLait), as required in the Grant Agreement for an Action with

Multiple Beneficiaries, number 50721.2013.001-2013.082. The report consists of a general

description of the progress during the first five months of the project as well as an update of

the metadata repository created by the predecessor to this project. In addition, there is a

separate note on deflators.

The main objectives of this Action are to further expand the methodology for ICT impact

analysis developed in the two earlier closely related projects (ICT Impacts and ESSLimit),

extend the time series to include data from more recent years for a group of European

countries, analyse how the economic environment influences the links between ICT and firm

performance and ready an infrastructure for future cross-country distributed research.1 Work

within the Action is carefully related to relevant policy areas such as the EU digital Agenda

and the MEETS programme for improving business statistics. The Action has a duration of

twelve months.

The project is organised into two workstreams (Administration and Impact Analysis) and

consists of partners from 14 European statistical offices supported by academic advisors.2

Statistics Sweden is the project leader, with Statistics Netherlands and Statistics Norway in

the steering group together with the sponsor Eurostat. As from May 2013, the Central

Statistical Office of Poland has been invited to join the steering group following a re-balance

of tasks provided by Statistics Norway.

According to plan, there has been a two-fold focus during the initial months of the project: on

creating good conditions for the analytical work and on starting or restarting the exploratory

or already ongoing analytical work. This means that issues relating to harmonisation of data,

data quality and development of code have been high on the agenda, as has been the forming

of internal analytical sub-groups in the project.

1ESSnet on Linking of Microdata on ICT Usage, Grant number 50701.2010.001-2010.578 and ICT Impacts

Grant number 49102.2005.017-2006.128. 2 The statistical offices of Austria, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, Poland,

Slovenia and the United Kingdom are all co-partners of the project.



3

Workstream (I) – Administration

Eva Hagsten, SCB

Objectives

Workstream I deals with general project management including data sharing and

dissemination of information, as listed in Table 1.

Table 1. Deliverables and Timing, Workstream I

Deliverables Timing

Provision of a secure infrastructure for data transfer and storage Lifetime of project

Provision of a website and project mailbox for internal and

external exchange of information and communication of results

Lifetime of project

[email protected], www.esslait.eu

Mid-term technical report End of month 5

Dissemination of datasets with detailed results End of project

Final report End of project

Financial report in accordance with conditions described in

the grant agreement 60 days after end of project

Note: Italic means remains to be done.

Achievements so far

The secure infrastructure, the project mailbox as well as the e-mail address were made

available from 1 January 2013. In addition, an outline of the project has been uploaded to the

EU Cros-portal (www.cros-portal.eu).

The project held its first steering meeting in Amsterdam on 7 February 2013, and focussed on

strategies. The launch workshop was held in Oslo on 21-22 March with an emphasis on data

and code. The Oslo meeting also kicked off the data quality work by holding a special

workshop on this theme.

The next steering meeting will take place in Stockholm on 18-19 June and the final

conference is planned for 15-16 October 2013 in Rome. The steering group maintains an

ongoing dialogue on project issues between meetings.

To guarantee a proper financial management of the project accounts, the project lead is

supported by a professional controller. A system for monthly follow-up of project costs has

been developed, including a time reporting sheet for each member country.

On April 8, the Grant Agreement was countersigned by Statistics Sweden. The advance

payment, 40 per cent of Eurostat funded costs, arrived during the second half of April and was

immediately distributed to the co-partners.

By the end of April, the project had spent slightly more than one-fifth of its total expected

costs, and somewhat more of the staff and travel budgets. This is lower than a proportionate

use of the total budget, but must be seen in the light of the first three months of the project

that called for inputs without a signed grant agreement. Project work also tends to be cyclical,

with periods of heavier input closer to deadlines.


http://www.esslait.eu/

http://www.cros-portal.eu/



4

Table 2. Budget and outcome

January-April 2013

Costs (Euro) Budget Outcome Proportion %

A.1) Costs of the staff assigned to the action 472371 115399 0.24

A.2) Travel and subsistence costs 56000 15384 0.27

A.3) Purchase cost of equipment 3000 244 0.08

A.4) Costs of consumables and supplies 0 0 0.00

A.5) Costs entailed by other implementation contracts 79698 24750 0.31

A.6) Any other direct costs 25000 6363 0.25

A.7) Indirect costs of implementation 141711 34620 0.24

TOTAL ELIGIBLE COSTS OF THE ACTION 777780 172560 0.22

TOTAL NON-ELIGIBLE COSTS 0 0 0.00

TOTAL COSTS OF THE ACTION 777780 172560 0.22

Looking at time spent across the project group, the variations are quite large, with the steering

group countries, as expected, together with the United Kingdom having put in the most time.

Graph 1. Time spent on project

January-April 2013, as proportion of budgeted days

Risks and mitigation

The main risk with this project arises from the issue of timely delivery of country outputs.

The project has a “hard” deadline, end of December 2013, which cannot be amended. Thus,

the project is dependent on timely deliveries of output so that analytical and data quality work

can proceed. The steering group and project lead have tried to make a head start in all project

areas, and have also planned ample time for each code run. Further risks with the project

relate to the quality of the data and dissemination of the output. These issues will be discussed

more in detail in the next section on impact analysis.

0%

5%

10%

15%

20%

25%

30%

35%

40%

AT DE DK FI FR IE IT LU NL NO PL SE SI UK TOT



5

Workstream (II) – Impact Analysis Michael Polder, Statistics Netherlands (CBS), Eva Hagsten, SCB, Patricia Kotnik, University

of Ljubljana

Objectives

The main task for this workstream is to update and further develop the project software

(Common Code) and methodology for impact and descriptive analyses initiated by the

ESSLimit and ICT Impacts projects. The undertaking also includes an update of the metadata

repository and the inclusion of more recent years to the project datasets. Furthermore, a

specific code will be created to allow future use of the data infrastructure. Table 3 presents the

expected deliverables from this workstream.

Table 3. Deliverables and Timing, Workstream II

Deliverables Timing

Report on updated metadata repository End of month 5

Common fixed code for reading, linking, output End of project

Updated description of project methodology and output, including

quality assessments (final report) End of project

Report on analytical results (final report) End of project

Release of output dataset End of project

Set of cross country indicators for public use End of project

Note: Italic means remains to be done.

Achievements so far

Most results from this workstream, except the metadata update, are not expected to be

finalised until the end of the project. However, the infrastructure for docking new modules to

the Common Code has already been tested.

During the process of sourcing input data, it became clear that detailed deflators for gross

output, intermediates and value added are either scarce or of low quality. This puts a certain

burden on not only the ESSLait project, but on all other research projects in need of data

below the aggregate level in pursuit of productivity measures. A separate note on deflators is

therefore added to the Interim Report.

Code and data work

After a thorough metadata update (presented in the next section of this report) it became clear

that the most recent year with full information available for all datasets and across project

countries is 2010. An old version of the Common Code was test run on these datasets

including 2010 before a revised version was released on 22 April.

The main changes to the code include:

Reading and construction of new variables, extracting results for further analysis

Addition of code to extract information on attrition and panel data

Addition of options in the program to save processed firm-level files



6

Cleaning of regressions to reduce running time

Development of two new code modules (DoDMComp and DoDynEmp, relating to the

analytical themes on ICT and competition and Reallocation and firm dynamics)

Extension of statistical output to support new research interests. For the

ICT/Innovation theme, the output was extended to include productivity statistics

separately for R&D and non-R&D performers, and by quartile of R&D expenditures.

The demography and industry dynamics files have been extended by age categories, in

accordance with ongoing work within the OECD Dynemp project.

Design of a new infrastructure, which allows additional modules to be docked to the

code

Two workshops were organized during the Oslo launch meeting to increase understanding

among project members of the code output and the possibilities to engage in code

development.

New variables

New variables were added to the list of project variables in close connection with the update

of metadata. Firstly, discussions with the project group led to the adjustment of the previous

definition of ADE (automated data exchange), where we now separate ADEGOV from the

rest of ADE categories, which are labelled ADENGOV. This is motivated by a suspected bias

in the original variable due to requirements of online reporting to authorities. Secondly, we

define a new set of variables (right-hand side variables refer to transmission format names of

the E-commerce survey):

INVOICE = max(INVREC, INVSND)

ADEINV = max(ADENGOV, INVOICE)

SISAPU = max(SISAINV, SISAACC, SISAPROD, SISADIST, SIPUINV, SIPUACC)

ECOM = max(AESELL, AEBUY)

INVOICE and ADEINV are variables that were introduced after discussion within the project

group, and reflect whether a firm automatically sends or receives invoices respectively, and

whether a firm uses ADE or automatic invoices.

The introduction of SISAPU is related to the fact that other e-business variables like ERP

(Enterprise resource management), CRM (Customer Relationship Management) and supply-

chain management, are missing for the year 2010. SISAPU is mainly meant to be an

alternative to the ERP variable, and relates to the sharing of information between various

business functions.

ECOM is a new dummy variable, which is the discrete counterpart of the existing variable

ECPCT (sum of e-sales and e-purchases), namely whether a firm makes use of e-sales or e-

purchases.

BIRTH =year of birth

BIRTH is the only other new variable added to the dataset. This follows from the uncertainty

of definitions of the previous AGE variable. The BIRTH variable should be defined in

accordance with the Eurostat firm demography concept, that is, a firm is born the first year it

has activities. With this change AGE will be calculated within the code.



7

Analytical themes

The previous round of the project (ESSLimit) included the following analytical themes:

I. ICT and human capital

II. ICT and innovation

III. E-business systems

IV. ICT, skills, and outsourcing

V. ICT and exports

VI. ICT and Resilience

VII. A new ICT intensity indicator

VIII. Descriptives

Since modules for these analyses are already present in the code, it was decided not to drop

any analytical themes at this stage. New tentative themes were presented and discussed at the

steering meeting in Amsterdam and at the launch workshop in Oslo.

Figure 1. Main Analytical Themes

Figure 1 illustrates all analytical topics, including presumptive new ones, grouped under five

main themes, as discussed at the project meeting in Oslo. The new themes are also described

under headings IX to XVI below.

IX ICT Expenditures

A recent study by the London School of Economics (LSE, Van Reenen et al. 2010),

shows that different forms of ICT capital have different effects on productivity. In

principle, it is possible to replicate and extend this study with data from the ICT

expenditures pilot survey. However, based on the metadata inquiry, the steering group

is reconsidering the possibilities for this theme, as there are only few countries with

New indicators

ICTi indicator

E-business

ICT expenditures

Dynamics

Resilience

Reallocation and firm dynamics

Adjustment costs

Intangibles

ICT and human capital

ICT and innovation

ICT, outsourcing

and skills

Quality and metadata

Metadata

Quality input data

Quality linked data

Selected topics

ICT and exports

ICT and employment

ICT and firm age

ICT and productivity

ICT and competition



8

panel data on the desired variables (NL, SE, and DK). This could take the character of

a country-specific analysis.

X. Reallocation

The OECD Dynemp project analyses employment dynamics by age and size classes,

based on micro-aggregated data from business registers. The ESSLait data enable

mimicking this research, and to incorporate ICT metrics to see if the employment

dynamics bear any relation to ICT. Such metrics have now been implemented in the

code. The steering group and certain project members are also actively cooperating

with the Dynemp research team at the OECD.

XI. Adjustment costs

This theme relates to the empirical evidence that ICT makes firms more flexible in

response to economic shocks. Ideally, gross job flows would be available from

employer-employee data. The steering group is reviewing the possibilities of this, as

well as collaboration with external parties on this theme.

XII. Quality of input data

This topic relates to issues that might affect the quality of input data and thus the level

of comparability of statistical indicators. The results of analyses across countries will

also be considered. For example, how do changes in the definitions and in the way

questions are asked in the ICT use questionnaire over the years affect the

comparability of the resulting statistics; what are the possible solutions to situations

where certain ICT use variables are not available in some years; and such issues as

consistency in coding of microdata, treatment of imputed values, and reliability of the

age variable. The theme is a joint effort by a sub-group of project members co-led by

Slovenia and Finland.

XIII. Quality of output data

The work on output data will examine issues arising due to linking of microdata sets,

with an emphasis on linking that can result in small sample overlaps. These overlaps

will be examined across countries and over time. The resulting biases will be studied

and illustrated by examining average values of indicators and correlation coefficients.

In addition, we will address the issue of how successfully does the process of ex-post

reweighting correct for these biases in the datasets. Measures that can be taken to

increase representativeness will be discussed. The theme is co-ordinated by Slovenian

Advisor.

XIV. ICT and employment

This subject will be pursued together with JRC/IPTS, which has formulated a research

proposal with some broad questions touching upon various relations between ICT and

employment. A special workshop will be organized in June to explore the possibilities

within this theme.

XV. ICT and firm age

An interesting question is whether the adoption of ICT differs by age group.

Currently, the steering group is reconsidering this theme, because it is still unclear

how the quality of the BIRTH variable can be assessed across countries.



9

XVI. ICT and productivity trends

This theme is a cooperative effort by some NSIs in the project group. It involves

research along the lines of a recent ONS publication, presenting descriptive

information on the relation between ICT and productivity in a cross-country setting,

with particular emphasis on dimensions otherwise not available and presumptive

effects of the financial crisis.

Since the project is conducted in the spirits of the EU Digital Agenda, Table 4 shows an

attempt to relate the project work to the seven pillars of this agenda. However, it is important

to bear in mind that the ESSLait and Digital Agenda goals do not fully coincide, thus these

links need to be considered intuitive rather than scientific. Broadly defined pillars are more

easily matched to the analytical themes.

Table 4. Analytical themes versus pillars of the Digital Agenda

Analytical themes (h)/ Pillar of the digital agenda (v)

ICTi in

dicato

r

E-bu

siness

ICT exp

end

itures

Resilien

ce

Reallo

cation

and

firm d

ynam

ics

Ad

justm

ent co

sts

ICT an

d h

um

an cap

ital

ICT an

d in

no

vation

ICT, o

utso

urcin

g and

skills

ICT an

d exp

orts

ICT an

d em

plo

ymen

t

ICT an

d firm

age

ICT an

d p

rod

uctivity

ICT an

d co

mp

etition

Digital single market x x x

Interoperability & standards x

Trust and security

Fast and ultra-fast internet access

x x x x

Research and innovation X x

Enhancing digital literacy, skills and inclusion

x x x x

ICT-enabled benefits for EU-society

x x x x x

Beyond the Digital Agenda, the analytical themes also interact with several other policy

relevant areas, as shown in table 5. These themes are much easier to relate to project work and

reveal an emphasis on impact and diffusion, while several other policy areas are still

represented to a less than negligible extent.



10

Table 5. Analytical themes versus policy areas.

Analytical themes (h)/ Policy areas (v)

ICTi in

dicato

r

E-bu

siness

ICT exp

end

itures

Resilien

ce

Reallo

cation

and

firm d

ynam

ics

Ad

justm

ent co

sts

ICT an

d h

um

an cap

ital

ICT an

d in

no

vation

ICT, o

utso

urcin

g and

skills

ICT an

d exp

orts

ICT an

d em

plo

ymen

t

ICT an

d firm

age

ICT an

d p

rod

uctivity

ICT an

d co

mp

etition

ICT policy: diffusion x x X x x x

ICT policy: impact x X x x x x x x

Structural economic policy: intangibles

x X x x x

Structural economic policy: business cycle

x x x

Structural economic policy: competition

x X

Labour and social policy: employment

x x x x

In addition to the analytical areas noted, the steering group is actively seeking cooperation

with other external parties, who may wish to make use of the existing data and research

infrastructure.

Conferences

Based on analytical work already started or ongoing from the previous round of the project,

papers were accepted for the NOEG conference in Innsbruck (May 2013, human capital),

ZEW ICT conference in Mannheim (June 2013, e-business) and the IARIW conference in

Sydney (November 2013, human capital). The steering group is also seeking the opportunity

to have analytical work peer reviewed by the CAED network during its upcoming 2013

conference in Atlanta, USA.

Furthermore, the project expertise has been sought by Framework 7 projects E-frame and

Mapcompete. The steering group has also engaged in an exchange of metadata experiences

with the project leader of the ESSnet on Data Warehousing, who was invited to the

Amsterdam steering meeting. Some of the changes in variables and code have been

undertaken with intensions to intensify collaboration with the OECD Dynemp research team.

Furthermore, the project is also cooperating with the ECB system project COMPNet.

Risks and Mitigation

As noted in the introduction, the major risk to the project is non-delivery of country outputs.

However, quality and dissemination dimensions could also be added to this. The three major

risks are listed below together with suggested measures of mitigation.



11

A. Non-delivery of country output

The main risk to the analytical work is that countries do not manage to deliver their output

datasets on time. Measures to mitigate these risks include:

Clear and reasonable deadlines for delivery.

Repeated reminders of deadlines.

Continuous support in code running by academic lead, project lead, and analytical

lead.

Data delivery is formulated as a contractual obligation.

B. Non-acceptance of data dissemination after lifetime of project

This risk threatens any analyses to be carried out after the project lifetime with the cross-

country dataset containing possible confidential data. The mitigation measures taken

comprise:

A distinction between cross-country dataset with public data (after disclosure controls)

and confidential data. The former dataset can always be published and made available

after the project lifetime.

External access to the project dataset under the consent of the project group, and after

signing a confidentiality agreement.

C. Non-reliable data

The quality of the research is subject to the consistency and quality of the underlying

microdata and the representativeness of the cross-country dataset. Several measures have been

taken to ensure this:

Quality of input and output data is defined as separate analytical themes.

Cobweb analyses to visualize representativeness of subsamples.

Test runs of the old code before launch of the new one.

Workshop on input data quality, including suggestions for improvements.

Workshop on interpretation and quality of the output data.

Detailed descriptions of output datasets and guidelines for quality assurance published

on the project website.

Preliminary quality control after each round of uploads performed by a set of

summary graphs that will identify aberrations in the data at the level of the total

economy in each project variable for each county.

Guidelines for disclosure control made available to limit the need of blanking out

cells.



12

A Note on Deflators Eric Bartelsman, Vrije Universiteit, Amsterdam

This section of the midterm report discusses the need for price deflator time series at the

country and industry level for gross output, intermediate purchases and value added. Next, it

discusses the sources of such deflators, along with problems of availability, industry

definition and methodology for each of the sources. Finally, the paragraph sketches the steps

that NSIs will need to take in order to meet the demands for deflators by the user community.

The main goal of the ESSLait project is to understand the interrelation between firm

performance and ICT use. To this end, the project needs to generate comparative performance

measures across countries, industries, and time that are based, in part, on real output at the

firm level. In addition to ESSLait, many other policy and academic research projects need

indicators based upon real output at the firm level. For example, productivity is one such

measure considered by researchers for international organisations such as the World Bank and

the OECD, or for policy work at the EU or member governments. Other relevant measures are

competitiveness (e.g. considered by the ECB COMPNET project, or the FP7-funded

MapCompete project) or real output of value chains (considered by e.g. the OECD).

All of these policy relevant measures have in common that they, ideally, make use of firm-

level price deflators for output, purchased inputs, and value added. However, in general, no

firm-level deflators are available.3 Instead, most firm-level research has made use of industry-

level deflators. Deflating firm-level data with industry deflators allows one to make meso-

level performance aggregates that are comparable to measures available in industry databases

such as EUKLEMS or STAN. Within industries, firm-level performance measures are not

truly ‘real’. For this reason, firm-level Total Factor Productivity measures recently have been

referred to as ‘Revenue TFP (TFPR)’ in the academic literature to distinguish them from

‘Quantity TFP (TFPQ)’, which is based on the volume of firm-level output. While the frontier

of academic research currently is concerned with the theoretical deviations between TFPR

and TFPQ as well as with empirical methods to bridge the gap through estimation of mark-

ups, ESSLait would be pleased if the required deflators were available to make TFPR-style

performance measures.

Given that NSIs have well developed programs to publish statistics on consumer and producer

prices, as well as GDP deflators, all the underlying data to publish good industry deflator time

series are available in-house. It is thus surprising that the research projects noted above have

difficulty finding deflators for the appropriate time periods, industries and value concepts.

Indeed, for the ESSLait project, we need deflators for gross output, intermediate purchases,

and implicitly for value added, for roughly 2- or 3-digit Nace 1.1 industries (A30) for the

period 2000-2010 (or starting in the 1990s for the long panels). For previous ESSnet projects

on ICT and performance, we made use of the deflators compiled by the EUKLEMS project,

which run through 2006 or 2007 for the A60 NACE 1.1 activities. We explored the following

sources to find deflators through 2010.

3 Sometimes the unit-value data at the firm-product level of detail from Prodcom could be used to make firm-

specific output and purchased materials deflators (for manufacturing inputs and outputs, for the largest firms).



13

1) EUKLEMS, on NACE 1.1 (A30) through 2007 (2006 in some countries); on NACE 2

(A31) for 2008-2010 or 2011 for six of the ESSLait countries, backcast using NACE1.1

growth rates. For some countries, NACE 1.1 deflators were available for A60-activities (2-/3-

digit).

2) WIOD on NACE 1.1 through 2009, (A30).

3) STAN on ISIC rev3 through 2008; on ISIC rev4 for 2009-10. 2-digit (roughly 30

industries).

4) Eurostat (via Transmission Programme from NSIs): Implicit annual price relatives for

Value Added from National Accounts, NACE2 (A64), 1994-2010.

At present, industry deflators are not a standard deliverable of NSIs through the Eurostat ESA

transmission programme. Instead, many NSIs generate industry-level supply and use tables at

current prices and at prices of previous year. Based on these data, we can chain the annual

implicit price relatives that are computed by dividing current price value added by VA in

prices of previous year. Unfortunately, there are gaps in the data. For example, the UK does

not conform to international standards in estimating real GDP, and thus does not have annual

supply and use tables on prices of previous year.

Besides lacking deflators for gross output and intermediates, the data from Eurostat are quite

‘noisy’. Even when ignoring obvious data errors (industry prices doubling or cut in half in a

given year), the data remain noisy. Limiting the observations to the calm period 2002-2006,

excluding agriculture, mining, and petroleum refining industries, and excluding the rapidly

changing transition economies, 10 percent of the observations have annual price relatives

larger than 1.12 or smaller than 0.91. By contrast, the EUKLEMS dataset barely has outliers,

and for the same sample as noted above, 10 percent of observations lie outside the 1.09 to .92

range.

Graph 2 shows a scatter plot of the price relatives from EUKLEMS (labelled dpw_va) and

from Eurostat (labelled dp_va) for the subsample noted above. Because the industry

classification differs between the two datasets, the industries are limited to those 2-digit

industries whose description has not changed and where underlying concordances are close to

‘one-to-one’ (see below). While it is not a priori clear which of the two deflators is ‘better’,

the differences between them (10 percent per year is not uncommon) overshadow the

magnitude of performance growth rates that we are trying to compare and contrast (1

percentage point extra growth in a year would please most policy makers).

The deflators from the STAN datasets are potentially useful. They are available for gross

output and intermediates as well as for value added. Because these data have been widely

used for more than a decade, most of the errors have been flagged and corrected. Furthermore,

staff efforts at the OECD ensure consistency and comparability of results across countries.

Unfortunately, the OECD stopped updating the ISIC rev3 version of the data in 2008/09,

making the deflators less useful for the ESSLait. In addition, the industry breakdown in

STAN differs from the one used in ESSLait.



14

Graph 2. Scatter plot of growth of VA price deflator, Eurostat versus EUKLEMS

The WIOD dataset contains many of the indicators of the EUKLEMS system, updated

through 2009. It should be noted that much of the updating used information available

through the STAN dataset. This dataset is available on NACE 1.1 at the A30 (2-digit) level of

detail. For the period prior to 2007, the deflators are very similar, though not identical, to

those published in the EUKLEMS system. For the current run of the common code (v4.0),

ESSLait extends the EUKLEMS deflators used in previous rounds of the project beyond 2006

using the WIOD deflators.

Discussion

Two methodological issues play a role when comparing the deflators. First, the conceptual

methodology which uses the underlying goods and services price indexes to generate

industry-level deflators. Second, the role played by industry classifications and their change

over time (NACE1.1 to NACE2).

Starting with the methodology, I present a hypothesis about a source of error, formed after

casual inspection of some underlying data. In some of the countries, the SUT system is

balanced each year, first in current prices, and then in prices of the previous year (ppy). In the

balancing, magnitudes are shifted across cells of the tables. The overall implicit price relative

of aggregate value added, given by the ratio of current to ppy for total GDP, stays accurate.

However, the implied price relative for total supply of a particular activity or the total use

intermediates of a particular activity may be different from what we would obtain by

computing a product sales-weighted or expenditures-weighted average of product deflators

for each activity. When chaining the implicit prices relatives obtained from balanced supply



15

and use tables, the annual fluctuations may become unrelated to the fluctuations of the

underlying price indexes.

The next puzzling bit of methodology relates to the industry classification changes. Clearly,

the structure of the economy changes over time. New products are introduced, old products

disappear and existing products are modified. Similarly, firms enter, firms exit, and

continuing firms change their inputs, outputs, and production methods. However, these

changes need not create discrete breaks in statistical time series, especially for industry-level

price deflators.4

Using underlying source data on prices available at NSIs will allow us to make good industry

deflators series for any industry classification. There are a few main sources of price series:

the CPI program, PPI program, Prodcom, trade data (with values and units of imports), and

‘special projects’ (e.g. for owner-occupied housing, financial services, health services).

Abstracting from details of treatment of taxes and subsidies, we could link each price series to

the appropriate product from a product classification (CPA). Next, the product classification

needs to be linked to the products from the supply and use tables. Finally, the industries

(activities) of these tables need to be linked to the industry classification of choice. With

enough disaggregation, this should be a simple concordance. Alternatively, we could use the

micro-level Prodcom table linked to firm-level output data, instead of the industry-by-product

SUTs, for manufactured products. With the appropriate industry label attached to each firm,

we could generate product-sales weighted output deflators or product-use weighted

intermediate goods deflators for each industry. Value-added deflators would then implicitly

be derived using the double-deflation method.

4 For firm-level aggregates, the ESSLait project describes how consistent times series of firm-level aggregates

can be generated for multiple classification hierarchies, as long as individual firms can be labeled with an

activity from a node of each hierarchy.



16

ESSLait Metadata Repository Ekaterina Denisova, Statistics Norway (SSB)

Objectives

The ESSLimit project built an extensive metadata repository on the sources underlying the

microdata linking. A description is presented here of how the repository has been updated for

the purpose of its successor, the ESSLait project, encompassing a discussion on possible

additional variables and how to achieve reliable and comparable information. There is also a

dialogue on how to deal with variables that change their definitions over time or suddenly

disappear from the questionnaire (or become biannual).

Introduction The present report summarizes the results of metadata collection activities in the framework

of the ESSLait project. The continuation of the project made it possible to extend the time

series by adding recent data for 2010.

Metadata surveys in ESSLait

The ESSLait metadata collection started in January 2013 with the distribution of a mini-

questionnaire on general availability of data for 2010-2011. Later on, three detailed

questionnaires on the Production Survey (PS), the Community Survey on ICT Usage and e-

Commerce in Enterprises (EC) and the Community Innovation Survey (CIS) were distributed

to and returned by the project group.

From ESSLimit to ESSLait

1. All the members of the ESSLimit consortium except Romania continued their engagement

with the project now in the shape of the ESSLait. The present update is based on PS and EC

metadata surveys provided by the whole project group and CIS data for 13 project members.

In Germany, this survey is outsourced to Zentrum für Europäische Wirtschaftsforschung

(ZEW).

2. Feedback from the project group indicated that a number of improvements were needed to

the metadata collection, in particular, to the design of the questionnaires. Nonetheless, it was

decided to leave the design of the questionnaires unchanged. The rationale behind this

approach was to ensure comparability of the answers for the whole period of the ESSLimit

and ESSLait projects: 1999-2010. The questionnaires on the EC and CIS contain a section

with generic information and sections with variable-specific questions, similar to the

questionnaires from ESSLimit. In addition, the structure of the PS remains the same: a

ReadMe section, a section on metadata guidance and variable-specific questions.

3. It became obvious from answers to the mini-questionnaire that an update of the PS with

data for 2011 was not feasible. The survey includes several variables from the Structural

Business Statistics (SBS) which is normally not released by most NSIs until May-June year

t+2. Hence, the period of time is expanded to 2010.

ICT Usage and e-Commerce in Enterprises (EC) As noted above, the EC metadata questionnaire was submitted by all 14 members of the

project. In addition to the standard generic questions (see Table 7), the survey collected

information about EC variables in 2010 and 2011.Moreover, it was decided to check



17

availability of data on ICT investments, expenditures and own software after discussions

within the project. Currently, inclusion of these is under discussion.

The summary of the answers to the EC metadata questionnaire is presented in Table 6. The

periods when these data were collected are included in the columns for ICT expenditure/

investment and own software dummies. A positive answer means that NSIs collected at least

some variables on ICT expenditure/ investment and own software. Furthermore, it can be seen

that several countries have information on ICT expenditures or investments, but unfortunately

only as time series in a sub-group of countries.

Table 6. Overview of Responses to EC Metadata Questionnaire

NSI EC 2010 ICT expenditure/ investment Own software from EC

AT Y N N

DE Y N N

DK Y Y, 2003-2011 Y, 2003-2011

FI Y N Y, 2006-2011

FR Y Y, 2010 N

IE Y N Y, 2004-2005

IT Y N N

LU Y N N

NL Y Y, 2009-2011 Y, 2000-2011

NO Y Y, 2009-2011 Y, 2004-2007

PL Y Y, 2009-2011 Y, 2003-2009

SE Y Y, 2005-2011 Y, 2002-2011

SI Y Y 2009, 2011 Y 2009, 2011

UK Y N N

2.1. Variables from EC 2010

The project members reported no deviations in the collection of the mandatory variables, but

some variations in the collection of optional variables. A complete list of EC 2010 variables is

available in Appendix 1, Table I.

One of the early findings of the ESSLait project was that the EC regularly undergoes

extensive changes in contents, most recently in 2010 and 2011 where variables either received

new definitions or were dropped altogether. Dealing with these kinds of changes has been a

challenge in the project work. The steering group assessed several alternative remedies,

including interpolation of 2010 values by using variables for 2009 and 2011. That is why the

EC metadata questionnaire was extended to 2011. In later stages the idea of interpolation of

2010 variables was rejected mainly due to comparability considerations.



18

Table 7. Generic Section from the EC Metadata Survey: General Survey Information Year

Variables Further Info

Number of observations Please state the number of valid survey responses for the particular survey

for which data will be used

Sample size Please give the total sample size of the survey, i.e., the total number of

questionnaires sent out

Survey Frequency How frequent is the question linked to this variable?

'End of Reference Period' (e.g. 31/12/2008)

If the survey assumes a point in time, when is this? For example in Austria the e-commerce survey refers to January of the survey year (Jan T)

Financial data reference period (e.g. January-December Year T)

If financial data is being surveyed, when is the reference date? For example in Denmark the e-commerce survey refers to the year previous to the survey

year (T - 1)

Level of firm surveyed (e.g. Enterprise Level, Reporting Unit Level or Local Unit

Level)? What sampling frame is used, i.e. from what source was the sample drawn?

In most cases this has been the Business Register

Sampling Frame (e.g. Business Register)

At what level is this survey aimed? This is particularly relevant for large enterprises where one business entity may receive more than one survey. Firm level could be 'Enterprise Level', 'Reporting Unit Level' or 'Local Unit

Level'. For example the U.K.’s e-commerce survey is aimed at the 'Enterprise' level.

Scope of survey, by economic activity: sections C, D, E, F, G, H, I, J, L, N, divisions

69-74, groups 95.1, 65.1, 65.2, classes 64.19, 64.92, 66.12, 66.19 Indicate if some of these industries are not covered

Scope of survey, by enterprise size: enterprises with 10 employees or more Indicate if there are deviations/ smaller enterprises are covered in addition

Industry Classification(e.g. SIC, NACE) What type of industry classification is used to in the dataset (e.g. sic92,

sic03, sic07, nace)?

Weighting scheme How is the data weighted to whole economy level? E.g using a-weights, g-

weights a combination of weights?

A review of the Eurostat model questionnaires for 2009 and 2010 shows that the variables can

be divided into three groups:

a) Variables with no changes compared to 2009,

b) Variables that were dropped in 2010 and

c) Variables that are completely new compared to 2009.

This report goes through the parts of EC 2010 that suffered changes. Variables from groups b)

and c) are thus presented in detail.

1. Intranet

The following variables on the use of intranet were excluded from EC 2010. The time series

for these variables extends to 2009.

ESSLait name Status Description

INTRA Dropped Firm has intranet

EMPINTRAPCT Dropped % of workers with access to intranet



19

2. Automated data exchange (ADE)

The first four variables were dropped from the EC 2010. The last two variables appear in EC

2010 for the first time. Comparability tests carried out by Eurostat shows that the use of

INVSNDAP and INVRECAP as proxies for INSND and INVREC was not justified.


ADESU Dropped Use of ADE for sending orders to suppliers

INVREC Dropped Use of ADE for receiving e-invoices

INVSND Dropped Use of ADE for sending e-invoices

ADECU Dropped Use of ADE for receiving orders

INVSNDAP New in 2010 Sending e-invoices in a standard structure suitable for automatic processing

INVRECAP New in 2011 Receiving e-invoices in a standard structure suitable for automatic processing

Based on discussions during the Amsterdam Steering and the Oslo launch meetings, new

variables for ADE were introduced to the project analysis: ADENGOV and ADEGOV. The

former was created for the purpose of the project and is derived from three underlying

variables: ADEPAY, ADEINFO and ADETDOC. All these variables as well as ADEGOV

have been surveyed since 2007. At the time of this report, it was unclear if all the project

members could include ADEPAY and ADEGOV in their input datasets for 2008 and 2009 as

both variables were still optional.

ESSLait name Description

ADEPAY Use of ADE for sending payment instructions to financial institutions

ADEINFO Use of ADE for sending product information

ADETDOC Use of ADE for sending transport documents

ADEGOV Use of ADE for sending or receiving data to/from public authorities

ADENGOV =(ADEPAY = 1) or (ADETDOC = 1) or (ADEINFO = 1)

3. Supply Chain Management (SCM), Enterprise Resource Planning (ERP) and Client

Relationship Management (CRM) software

These are the major e-business systems variables that suddenly disappear from the production

year 2010. No proxies for SCM and CRM have been found. The time series for these

variables end in 2009. SISAPU has been suggested as an alternative to ITERP (see paragraph

“New variables” in the report on Workstream (II) – Impact Analysis).


SISU Dropped Sharing Supply Chain Management (SCM) data with suppliers

SICU Dropped Sharing SCM data with customers

ITERP Dropped Enterprise Resource Planning

CRMSTR Dropped Use of CRM software to share of information with other business functions

CRMAN Dropped Use of CRM software to analyse information for marketing purposes

4. E-commerce variables

In EC 2010, the variable AESELL on sales through the internet (or EDI) was split in two:

sales via a website or sales via EDI-type messages. Hence, for 2010 it was derived from

AWSELL and AXSELL. The same holds for AESVALPCT. The split of AEBUY, firms’

orders through internet, was optional in EC 2010. Half of the project members chose to

include optional questions on AWBUY and AXBUY, while the other half stuck to the 2009

question on AEBUY. Apart from these splits, there were no changes in the formulations of



20

the e-commerce questions in EC 2010, implying that it is possible to keep the same set of e-

commerce variables as in the ESSLimit project.

ESSLait Status Description

AESELL Dropped Firm sells through internet (or EDI)

AESVALPCT Dropped % of sales through internet (or EDI)

AWBUY* New in 2010 Firm orders via a website

AXBUY* New in 2010 Firm orders via EDI-TYPE messages

AWSELL New in 2010 Firm sells via a website

AXSELL New in 2010 Firm sells via EDI-type messages

AWSVALPCT New in 2010 % of sales via a website

AXSVALPCT New in 2010 % of sales via EDI-type messages

ICT expenditures/investments

Data on ICT expenditure/ investment have been optional to survey so far. Some project

members have experience in collecting these variables, for example through participation in

pilot studies of ICT expenditures/ investments conducted by Eurostat for production years

2004 and 2009. Other members have been conducting independent ICT expenditure/

investment surveys for quite a while, for example Denmark since 2003 and Sweden since

2005. Data collection tools may vary across NSIs: EC, SBS or other surveys.

Questions on ICT Investments/Expenditure are presented in Appendix 1, table II. They

originate from the Eurostat Methodological Manual for surveys on ICT

Investment/Expenditure, production year 2009. The formulation of these questions in the

surveys before this year may deviate from the 2009 model, though quite a few remain

comparable over time. Inclusion of ICT expenditure/investment in the analysis remains an

open question.

Optional questions on own software from EC surveys

Another possible data source for ICT intangibles- own software- is the section Additional

questions introduced in the national questionnaire in the EC surveys for the years 2004-2008.

From 2008 onwards, these questions are no longer included in the Eurostat quality report

templates. Nevertheless, some countries continue to collect these data. The list of the

variables on own software is available in Appendix 1, table III. The possible inclusion of

these variables in the project analysis is under discussion.

Production Survey (PS) The Production Survey contains a number of firm-level economic variables. Some of them are

collected via the SBS; others come from different sources including the Business Register, the

education statistics (HK-variables) and the trade surveys (EXPORT and NX). The list of PS

variables in the metadata survey has been kept unchanged since ESSLimit, with the exception

of two new questions: the link between Business Register and Employer- Employee Register

and the possibility of extending long panel data set from the ESSLimit project. Tables 8 and 9

summarize the responses. Table V in Appendix 1 presents a more detailed description of the

PS variables. Table IV contains a list of questions from the variable-specific section of the PS

metadata survey.



21

Table 8. Overview of the Answers to PS 2010 Metadata Survey ESSLait name Description AT DE DK FI FR IE IT LU NL NO PL SE SI UK

NV nominal value added (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y Y

NQ nominal gross output (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y Y

E full-time employment Y Y Y Y Y Y Y Y Y Y Y Y Y Y

PAY total wage bill (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y Y

NM nominal expenditures on intermediates (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y Y

K capital services measure (in national currency) Y Y Y Y Y Y Y N Y Y Y Y N Y

HKPCT pct workers with post upper secondary education Y N Y Y N Y N N N Y N Y Y Y

HKITPCT pct workers with post upper secondary IT education Y N Y Y N Y N N N Y N Y N Y

HKNITPCT pct workers with post upper secondary non-IT education Y N Y Y N Y N N N Y N Y N Y

EXPORT firm exports of goods and services Y Y Y Y Y Y Y Y Y Y Y Y Y Y

NX firm exports of goods and services (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y Y

Wgt_PS sample weight on business survey Y N N N N N N N Y N N N N Y

EMP_BR number of employees given in Business Register Y Y Y Y Y Y Y Y Y Y Y Y Y Y

FRGN_OWN dummy for foreign ownership Y Y Y Y Y Y N N Y Y Y Y Y Y

MNC dummy for multinational corporation Y N Y N Y N N N Y N Y Y Y N

AGE age of firm in given year Y N Y Y Y Y Y Y Y Y Y Y Y Y

When the answers to the PS metadata survey had already been submitted, it was decided to

replace AGE by BIRTH, as described in the section on Impact Analysis. This is the only

disparity between the list of the variables in the code and the PS metadata survey. This change

follows from the need for harmonising the AGE variable, which could be done by using the

Eurostat firm demography definition for year of birth (first year of economic activity) and

deriving the AGE variable from this.

Table 9. Overview of the Answers to PS 2010 Metadata Survey

NSI

Link between Business Register and Employer-Employee Register

Extend long panel data set

AT N N

DE N N

DK Y N

FI Y Y

FR Y Y

IE N Y

IT N N

LU Y N

NL Y Y

NO Y N

PL Y N

SE Y Y

SI Y Y

UK N Y



22

The review of the answers to the PS metadata survey did not indicate any significant changes

in PS variables since the previous round. The observations in the ESSLimit Interim Report on

Metadata Collection hold true for the 2010 metadata as well. It seems appropriate to only

mention information which is new compared to ESSLimit user guide.5. It mainly concerns

data sources for three variables: dummies for foreign ownership and multinationals and value

of exports. Table 10 contains updates for them. The design of the table follows the ESSLimit

presentation, with an additional row ESSLait Update, featuring new information.

Table 10. PS Metadata Update for FRGN_OWN, MNC and EXPORT INTENSITY FRGN_OWN Dummy for Foreign Ownership

Definition Foreign ownership refers to whether a firm is owned and controlled by a foreign entity.

Required Format Boolean

Required range 0 – 1

How to derive (from ESSLIMIT)

i. This can be derived from firm level Foreign Direct Investment (FDI) data where information on inward FDI can be obtained. ii. FATS (Foreign Affiliates Statistics). Norway and Romania use this source. Eurostat has high level Inward FATS data for most countries; as such it might be worth searching your NSI for the source data at firm level. iii. Trade data could also be a resource.

ESSLAIT Update iv. NL reported CIS as a data source for this variable.

MNC Dummy if firm is multinational

Definition A multinational here refers to a firm that engages in FDI. As such a foreign owned firm (i.e. FRGN_OWN=1) is a multinational, but not vice-versa. E.g. ESSO UK (foreign owned/multinational) vs BP (not foreign owned/multinational).

Required Format Boolean


How to derive (from ESSLimit)

i. FATS (Foreign Affiliates Statistics) data can also be helpful here. Eurostat website shows that there is data on Inward and Outward FATS. From our definition above, it can be inferred that while foreign ownership relates to Inward FATS, Multinational relates to firms either Inward or Outward FATS, (i.e., Inward: FRGN_OWN=1 & MNC=1; Outward: MNC=1). ii. Trade data might also be a resource.

ESSLait Update iii. NL reported CIS as a data source for this variable. iv. PL used SBS as a data source. v. SI derived MNC from FDI.

Export intensity Export values relative to sales

Definition What share of sales is exported

Required Format Pct


How to derive (from ESSLimit)

This could be derived from trade surveys, business statistics or VAT registers with questions on export or export orders.

ESSLait Update ii. SBS is used as a data source by LU and UK (UK only for services)

5 See Eurostat (2012), Annex 3 in Annexes to the final report in ESSnet on Linking of Microdata on ICT Usage



23

Community Innovation Survey (CIS)

In the design of the CIS 2010 metadata survey, the census approach from the ESSLimit

project was followed. This allows the ESSLait project to obtain an overall picture of CIS

variables for 2010 and enables the inclusion of other relevant variables in the analysis later

on, during the course of the project.

The CIS 2010 metadata questionnaire is largely based upon the 2008 metadata survey

designed in ESSLimit. In addition to variables that are surveyed from both ways, completely

new variables from CIS 2010 are included. A change from ESSLimit is introduced in the

variable descriptions. The earlier short descriptions are now replaced by the question text

from the Guidelines for the Transmission of National Data6.

A comparison of the CIS 2008 and CIS 2010 model questionnaires reveals three types of

changes7:

a) Variables are collected in both waves, but have differently formulated questions that

affect comparability: INPDTW, INPCSW

b) Variables that were dropped in CIS 2010: MKTMET

c) Completely new variables compared to CIS 2008: INPDFC, INPDFE, INPDFW, as

well as a module on skills, EMPUD.

Variables with changed definitions

Questions about who developed product and process innovations are present in both

CIS 2008 and CIS 2010; but the way the questions are formulated make them

incomparable. First, in CIS 2010 a split in goods and services is introduced. Second, in

2010 firms are suddenly free to tick several options, while previously they were asked

to choose the most representative answer. Hence, it was not feasible to derive

comparable values for INPDTW and INPSCW from 2010 data, and these variables

were excluded from the code.

Extract from CIS 2008:

6 See CIS 2010 General Data Structure for Transmission to Eurostat

7 For the complete list of CIS 2010 variables consult Appendix 1, Table VI.



24

Extract from CIS 2010:

Variables that were dropped

MKTMET was also dropped from the analysis.

New variables

The newly added variables are presented in Table 11 below. In general, NSIs’ flexibility with

regard to what questions to include in their national questionnaires may restrict the scope of

analysis in a projects like this when a group of several NSIs must agree on a set of variables

which remain stable from one survey year to another. The present summary shows that none

of the newly added variables is collected by the whole project group.

Table 11. Completely New Variables in CIS 2010 Eurostat name Question AT DK FI FR IE IT LU NL NO PL SE SI UK

INPDFC Were any of your product innovations during 2008-2010 first in your country Y N Y Y N Y Y N Y Y Y Y N

INPDFE Were any of your product innovations during 2008-2010 first in Europe Y N Y Y N Y Y N Y Y Y Y N

INPDFW Were any of your product innovations during 2008-2010 world first Y Y Y Y N Y Y N Y Y Y Y N

SGALA Graphic arts/ layout/ advertising Y N N Y Y Y N N N Y Y Y Y

SDOS Design of objects or services Y N N Y Y Y N N N Y Y Y Y

SMED Multimedia Y N N Y Y Y N N N Y Y Y Y

SWDS Web design Y N N Y Y Y N N N Y Y Y Y

SSWD Software development Y N N Y Y Y N N N Y Y Y Y

SMKR Market research Y N N Y Y Y N N N Y Y Y Y

SENAP Engineering / applied sciences Y N N Y Y Y N N N Y Y Y Y

SMSDM Mathematics / statistics / database management Y N N Y Y Y N N N Y Y Y Y

EMPUD % of employees in 2010 with a university degree Y N N N N Y Y Y Y Y N Y Y

Note: Several of the countries that did not include the question on educational achievement are able to derive this information from other sources.

Conclusion According to the project time schedule, the update of metadata is to be delivered several

months before the finalisation of the project. Therefore this version of the metadata update

can by no means foresee possible further needs during the course of analytical work.



25

Nevertheless, it covers the major changes in data for 2010. In case additional variables are

considered, they will be listed together with the final reporting of the project.

The metadata work has also drawn attention to the strong fluctuations over time in the design

of the EC questionnaire. The ICT field has developed quickly and partly unexpectedly;

however, it is desirable to allow variables to wear out instead of suddenly removing them or

making them biannual. Without a core of stable variables over time, it is impossible to

analyse trends. Mixing annual and biannual variables in the same survey is also something

that should be discouraged. As a comparison, the CIS is fully designed for biannual

responses, and as such, works well. The EC does not. Increased consistency in the datasets

over time would also go well in line with certain intentions of the MEETS programme, by

making better use of already existing data.



26

Appendix 1. Table I. Variables from EC2010 (Y- yes, N- no) Not collected in EC 2010

New variable in EC 2010 (Not collected in 2009)

* - optional question according to Eurostat regulation

Eurostat name AT DE DK FI FR IE IT LU NL NO PL SE SI UK

CUSE Y Y Y Y Y Y Y Y Y Y Y Y Y Y

EMPCUSEPCT* N Y N Y Y Y Y N Y Y Y N Y Y

INTRA Y

EMPINTRAPCT Y

IACC Y Y Y Y Y Y Y Y Y Y Y Y Y Y

DSL Y Y Y Y Y Y Y Y Y Y Y Y Y Y

BBOTH Y Y Y Y Y Y Y Y Y Y Y Y Y Y

MOBBB Y Y Y Y Y Y Y Y Y N Y Y Y N

MOBBBM* Y Y Y Y N Y N N Y Y Y N Y Y

MOBBBH* Y Y Y Y N Y N N Y Y Y N Y Y

MOBOTH Y Y Y Y Y Y Y Y Y Y Y Y Y Y

DIALUP Y Y Y Y Y Y Y Y Y Y Y Y Y Y

EMPIUSEPCT* N Y N Y Y Y Y Y Y Y Y Y Y Y

WEB Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADE Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADESU

INVREC

INVSND Y

ADECU

INVSNDAP Y Y Y Y Y Y Y Y Y Y Y Y Y Y

INVRECAP Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADEINFO Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADETDOC Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADEPAY Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ADEGOV Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SISU

SICU

SISAINV Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SISAACC Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SISAPROD Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SISADIST Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SIPUINV Y Y Y Y Y Y Y Y Y Y Y Y Y Y

SIPUACC Y Y Y Y Y Y Y Y Y Y Y Y Y Y

ITERP

CRMSTR

CRMAN

AEBUY Y Y Y Y Y Y Y Y Y Y Y Y Y Y

AWBUY* N N N Y N Y N N Y Y Y N Y Y

AXBUY* N N N Y N Y N N Y Y Y N Y Y

AEBVALPCT* N N N N Y Y N N Y N Y N Y Y

AEBCLS* Y N N Y Y N Y Y N Y Y Y N Y

AESELL Y

AESVALPCT Y Y

AWSELL Y Y Y Y Y Y Y Y Y Y Y Y Y Y

AXSELL Y Y Y Y Y Y Y Y Y Y Y Y Y Y

AWSVALPCT Y Y Y Y Y Y Y Y Y Y Y Y Y Y

AXSVALPCT Y Y Y Y Y Y Y Y Y Y Y Y Y Y



27

Table II. ICT expenditure/investment Have these figures ever been published? If yes, for which years?

Question nr. Text (Y=yes, N=no)

i1. Purchases of IT goods (computers and peripheral equipment) and Communication goods (equipment), during YEAR

a) Purchases of IT goods

b) Purchases of Communication goods

i2.

Share (percentage) of purchased IT goods (computers and peripheral equipment) and Communication goods (equipment) included in the balance sheet (investment), during YEAR

a) Purchases of IT goods included in the balance sheet (investment)

b) Purchases of Communication goods included in the balance sheet (investment)

i3. Purchases of other ICT goods (Consumer electronic equipment, Miscellaneous ICT components and goods, Manufacturing services for ICT equipment), during YEAR

i4. Share (percentage) of purchased other ICT goods (Consumer electronic equipment, Miscellaneous ICT components and goods, Manufacturing services for ICT equipment) included in the balance sheet (investment), during YEAR

i5. Purchases of software, pre-packaged and customised (Business and productivity software and licensing services), during YEAR

i6. Share (percentage) of purchased software, pre-packaged and customised (Business and productivity software and licensing services) included in the balance sheet (investment), during YEAR

i7a Labour input in own-account software, working hours or years FTE, during YEAR

i7. Total cost for creation of own account software, during YEAR

i8. Share (percentage) of total cost for creation of own account software included in the balance sheet (investment), during YEAR

i9. Purchases of Information technology consultancy and service, Telecommunication services) and other ICT services, during YEAR

i9a. Purchases of Telecommunication services, during YEAR

i10. Share (percentage) of purchased Information technology consultancy and services, Telecommunication services and other ICT services included in the balance sheet (investment), during YEAR

i10a.

Share of purchased Telecommunication services included in the balance sheet, during YEAR

i11. Purchases of Lease or rental services for ICT equipment, during YEAR

a) Purchases of operating lease or rental services for ICT equipment

b) Purchases of financial lease services for ICT equipment

i12. Share of purchased Lease or rental services for ICT equipment included in the balance sheet (investment), during YEAR

a) Share of purchased operating lease services for ICT equipment included in the balance sheet

b) Share of Purchased financial lease services for ICT equipment included in the balance sheet



28

Table III. Optional questions on own software in EC surveys

The questions originate from the Community Survey on ICT Usage and e-Commerce in Enterprises report templates for production years 2004-2008, section Additional questions introduced in the national questionnaire. From 2008, these questions were no longer included in the templates. Nevertheless, some countries might have kept them.

Question (Y=yes, N=no)

1. Does the enterprise have its own employees (exclude external consultants) who work with the development of software, e.g. system developers and programmers?

2. Estimate the number of man-labour years (a man-labour year is the work a full time employed person carries out in a year) for work on software development.

3. Distribute the number of man-labour years from the previous question on 3a. Software development for the enterprise’s own usage, e.g. programs for financial control or administration

3b. Software development for external sales (including software which forms a part of the products your enterprise sells)

3c. Other, e.g. maintenance, support, repair, etc.

Table IV. List of Questions from the Variable-Specific Section in PS Metadata Survey

ESSLait Variable Name List of variables which are relevant to the ICT impacts, ESSLimit and ESSLait projects. These are the same for all countries.

Project Variable Description Description of variables listed in previous column

Quote of survey question relating to variable (in English)

Please identify which question in the survey that the specified variable is linked with. Translated into English. If multiple choice, please include the options given.

Question number Please state the question number for the survey question quoted in the previous column.

Source Survey Name

What survey is the question linked to this variable found in? Please enter "derived variable" if not survey data, and provide a NOTE of how data are derived.

Format (and Units)

For example: Level, Percentage or Boolean. For Level and Percentage variables, please provide the UNITS. This is expected to be either count or currency, please confirm if units are in '000s, Millions (M), and the relevant currency (e.g. £, €). For Boolean variables, please confirm states represented by 0 and 1.

Range

Simply state the numerical range. For example, a Boolean variable might be 0 - 1. A percentage variable may be 0 - 100. Please indicate where level variables are non-positive.

Flow or stock?

Are data flows over periods of time, or snapshots at a point in time? (e.g. Turnover could be annual totals, while employment could be at a point in time).

Reference Period for Variable When is the reference period? For example "CY 2000", "YE 2001", "15/12/00". (CY= Calendar Year; YE= Year Ending)

Level of firm the survey is aimed at

At what level is this survey aimed? This is particularly relevant for large enterprises where one business entity may receive more than one survey. Firm level could be 'Enterprise Level', 'Reporting Unit Level' or 'Local Unit Level'. For example the U.K's eCommerce survey is aimed at the 'Enterprise' level.

Sampling Frame (e.g. Business Register) What sampling frame is used, i.e. from what source was the sample drawn? In most cases this has been the Business Register.

Additional metadata comments Please use this space to identify any additional metadata concerns or comments.



29

Table V. Guidance on PS Variables Variable

Name Variable Description Additional Notes

e Employment variable

Ideally, this refers to total full-time employees (including full-time equivalents for part-time workers). If full-time equivalents are not available, please give the closest figure that captures the total number of employees (head count) and indicate whether it is a point in time estimate or a period average.

Frgn_own Dummy for foreign ownership

This should be a 'yes' or 'no' to whether the firm is owned and controlled by a foreign entity.

K Capital stock measure

The capital stock measure refers to the stock (not flow) of productive capital which the firm has at the end of the referenced period. Might be survey based or could be derived. Survey estimates could include for instance book value or replacement value of assets such as structures, plant and machinery, vehicles, etc.

Mnc Dummy if firm is multinational

This should be a 'yes' or 'no' to whether the firm has operations in other countries. A multinational here refers to a firm which engages in or receives Foreign Direct Investment. As such a foreign owned firm is a multinational, but not vice-versa. E.g. ESSO UK (foreign owned/multinational) vs BP (not foreign owned/multinational).

Nm Intermediate purchases Refers to the value of goods, materials and services bought by the firm in the reference period

Nq Gross output / sales / turnover

What was the firm's Gross output/Sales/Turnover in the reference period?

Nv Value added

If a question on value added exists, please input the values, otherwise this can be derived by the formula (NQ-NM), i.e., Gross output - Intermediate Purchases.

Nx Exports of goods and services Refers to nominal value of exports

Pay Total employment expenditures or payroll What was the firm's total employment costs for the referenced period?

Wgt_PS Sample weight on business survey

Are weights available for the dataset and specifically can they be used to scale up the survey sample to the population total?

emp_br

Number of employees given in Business Register Number of employees given in Business Register

hkpct

Share of employees with post upper secondary education

By post upper secondary education, we generally mean university or similar education with the duration of at least three years. 'hkpct' refers to employees with post upper secondary education, and these are covered by: i. National code SUN2000Niva >= 53 (i e 53, 54, 62, 64), or ii. ISCED97 levels: 5A and 6.

hkitpct

Share of employees with post upper secondary IT education

‘hkitpct’ refers to employees with post upper secondary IT education, and are covered by: i. The following sub-groups within 'hk': National code SUN2000Inr - 440,441, 449, 44, 46, 48, 52, 54, 58, or ii. with-in 'hk', ISCED97 narrow fields 44, 46, 48 and broad fields 5.

hknitpct

Share of employees with non-IT post upper secondary education

‘hknit' refers to employees with post upper secondary non-IT education: These are covered by all other field definitions in the hk group apart from those in hkit.

export

Export dummy, exporter = 1 non-exporter = 0 Export dummy, exporter = 1 non-exporter = 0

age Age of firm in year Age of firm in year

Table VI. Community Innovation Survey 2010 Metadata Questionnaire Not collected in CIS 2010

New variable in CIS 2010 (not collected in CIS 2008)

Eurostat name Question AT DK FI FR IE IT LU NL NO PL SE SI UK

ENTGP Enterprise part of a group Y N Y Y Y Y Y Y Y Y Y Y N

HO Country of head office Y N Y Y Y Y Y Y Y Y Y Y N

MARLOC local/regional Y N N Y Y Y N Y Y Y Y Y Y

MARNAT National Y N N Y Y Y Y Y Y Y Y Y Y

MAREUR Other EU/EFTA/CC market Y N N Y Y Y Y Y Y Y Y Y Y

MAROTH All other countries Y N N Y Y Y Y Y Y Y Y Y Y

INPDGD Introduced onto the market a new or significantly improved good Y Y Y Y Y Y Y Y Y Y Y Y Y

INPDSV Introduced onto the market a new or significantly improved service Y Y Y Y Y Y Y Y Y Y Y Y Y

INPDTW Who mainly developed these products? Tick all that apply N Y Y N N Y N Y N Y N Y Y

INPDTG1 Goods innovations- Your enterprise by itself Y N Y N N Y Y Y Y Y Y Y Y

INPDTG2 Goods innovations- Your enterprise together with other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPDTG3 Goods innovations- Your enterprise by adapting or modifying goods or services originally developed by other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPDTG4 Goods innovations- Other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPDTS1 Service innovations- Your enterprise by itself Y N Y N N Y Y Y Y Y Y Y Y

INPDTS2 Service innovations- Your enterprise together with other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPDTS3 Service innovations- Your enterprise by adapting or modifying goods or services originally developed by other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPDTS4 Service innovations- Other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

NEWMKT Did the enterprise introduce a product new to the market Y Y Y Y Y Y Y Y Y Y Y Y Y

NEWFRM Did the enterprise introduce a product new to the firm Y Y Y Y Y Y Y Y Y Y Y Y Y

TURNMAR % of turnover in new or improved products introduced during 200-2010 that were new to the market Y Y Y Y Y Y Y Y Y Y Y Y Y

TURNIN % of turnover in new or improved products introduced during 200-2010 that were new to your firm Y Y Y Y Y Y Y Y Y Y Y Y Y

TURNUNG % of turnover in unchanged or marginally modified products during 2008-2010 Y Y Y Y Y Y Y Y Y Y Y Y Y



31


INPDFC Were any of your product innovations during 2008-2010 first in your country Y N Y Y N Y Y N Y Y Y Y N

INPDFE Were any of your product innovations during 2008-2010 first in Europe Y N Y Y N Y Y N Y Y Y Y N

INPDFW Were any of your product innovations during 2008-2010 world first Y Y Y Y N Y Y N Y Y Y Y N

INPSPD Introduced onto the market a new or significantly improved method of production Y Y Y Y Y Y Y Y Y Y Y Y Y

INPSLG Introduced onto the market a new or significantly improved logistic, delivery or distribution system Y Y Y Y Y Y Y Y Y Y Y Y N

INPSSU Introduced onto the market a new or significantly improved supporting activities Y Y Y Y Y Y Y Y Y Y Y Y N

INPCSW Who mainly developed these processes? Tick all that apply N Y Y N N Y N Y N Y N Y Y

INPSDV1 Process innovations- Your enterprise by itself Y N Y N N Y Y Y Y Y Y Y Y

INPSDV2 Process innovations- Your enterprise together with other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPSDV3 Process innovations- Your enterprise by adapting or modifying goods or services originally developed by other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y N

INPSDV4 Process innovations- Other enterprises or institutions Y N Y N N Y Y Y Y Y Y Y Y

INPSNM Were any of your process innovations introduced between 2008 and 2010 new to your market N N N Y N Y Y Y Y Y Y Y Y

INABA Abandoned or suspended before completion Y Y Y Y Y Y Y Y Y Y Y Y

INONG Still ongoing at the end of the 2010 Y Y Y Y Y Y Y Y Y Y Y Y

RRDIN Engagement in intramural R&D Y Y Y Y Y Y Y Y Y Y Y Y Y

RDENG Type of engagement in R&D Y N Y Y N Y Y Y Y Y Y Y N

RRDEX Engagement in extramural R&D Y Y Y Y Y Y Y Y Y Y Y Y Y

RMAC Engagement in acquisition of machinery Y Y Y Y Y Y Y Y Y Y Y Y Y

ROEK Engagement in acquisition of external knowledge Y Y Y Y Y Y Y Y Y Y Y Y Y

RTR Engagement in training for innovative activities Y Y Y Y N Y Y Y Y Y Y Y Y

RMAR Engagement in market introduction of innovation Y N Y Y N Y Y Y Y Y Y Y Y

RDSG Design Y N Y Y N Y Y Y Y Y Y Y Y

RPRE Engagement in other preparation Y N Y Y N Y Y Y Y Y Y Y N



32


RRDINX Expenditure in intramural R&D (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y

RRDEXX Purchase of extramural R&D (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y

RMACX Expenditure in acquisition of machinery (in national currency) Y Y Y Y Y Y Y Y Y Y Y Y Y

ROEKX Expenditure in acquisition of external knowledge Y Y Y Y Y Y Y Y Y Y Y Y Y

RTOT Total of these four innovation expenditure categories (in national currency) Y Y N Y Y Y Y Y Y Y Y Y N

FUNLOC Public funding from local or regional authorities N N Y Y N Y Y Y N Y N Y Y

FUNGMT Public funding from central government N N Y Y N Y Y Y N Y N Y Y

FUNEU Public funding from the EU N N Y Y N Y Y Y N Y N Y Y

FUNRTD Funding from EU's 6th or 7th Framework Programme for RTD N N Y Y N Y Y Y N Y N Y N

SENTG Sources from within the enterprise or enterprise group N Y Y Y N Y Y N Y Y Y Y Y

SSUP Sources from Suppliers of equipment, materials, etc. N Y Y Y N Y Y N Y Y Y Y Y

SCLI Sources from Clients or customers N Y Y Y N Y Y N Y Y Y Y Y

SCOM Sources from Competitors and other enterprises of same industry N Y Y Y N Y Y N Y Y Y Y Y

SINS Sources from consultants, commercial labs or private R&D institutes N Y Y Y N Y Y N Y Y Y Y Y

SUNI Sources from Universities or other higher education institutes N Y Y Y N Y Y N Y Y Y Y Y

SGMT Sources from Government or public research institutes N Y Y Y N Y Y N Y Y Y Y Y

SCON Sources from professional conferences, trade fairs, meetings N Y Y Y N Y Y N Y Y Y Y Y

SJOU Sources from Scientific journals, trade/scientific publications N Y Y Y N Y Y N Y Y Y Y Y

SPRO Sources from Professional and industry associations N Y Y Y N Y Y N Y Y Y Y Y

CO Cooperation arrangements on innovation activities Y Y Y Y Y Y Y Y Y Y Y Y N

PMOS Most important co-operation partner Y N N Y Y Y Y Y Y Y Y Y N

ORANGE Increased range of goods or services Y Y N Y Y Y Y N Y Y Y Y Y

OREPL Replace outdated products or processes Y Y N Y Y Y Y N Y Y Y Y Y

ONMOMS Enter new markets or increase market share Y Y N Y Y Y Y N Y Y Y Y Y

OQUA Improve quality of goods or services Y Y N Y Y Y Y N Y Y Y Y Y

OFLEX Improve flexibility for producing goods or services Y Y N Y Y Y Y N Y Y Y Y Y

OCAP Increase capacity for producing goods or services Y Y N Y Y Y Y N Y Y Y Y Y

OLBR Reduce labour costs per unit output Y Y N Y Y Y Y N Y Y Y Y N

OHESY Improve health or safety of your employees Y Y N Y Y Y Y N Y Y Y Y Y

ORGBUP New business practices for organising work or procedures Y Y Y Y Y Y Y Y Y Y Y Y Y



33


ORORED Reduce time to respond to customer or supplier needs Y Y N Y Y Y Y N Y Y Y Y N

ORGWKP New methods of workplace organisation Y Y Y Y Y Y Y Y Y Y Y Y N

ORGEXR New methods of organising external relations Y Y Y Y Y Y Y Y Y Y Y Y Y

OROABL Improve ability to develop new products or processes Y Y N Y Y Y Y N Y Y Y Y N

OROQUA Improve quality of your goods or services Y Y N Y Y Y Y N Y Y Y Y Y

ORORCO Reduce costs per unit output Y Y N Y Y Y Y N Y Y Y Y Y

OROCIN Improved communication or information sharing Y Y N Y Y Y Y N Y Y Y Y N

MKTDGP Significant changes to the aesthetic design or packaging Y Y Y Y Y Y Y Y Y Y Y Y N

MKTPDP New media or techniques for product promotion Y Y Y Y Y Y Y Y Y Y Y Y N

MKTPDL New methods for product placement or sales channels Y Y Y Y Y Y Y Y Y Y Y Y N

MKTPRI New methods of pricing goods or services Y Y Y Y Y Y Y Y Y Y Y Y N

MKTMET 2004-06 New or significantly changed sales or distribution methods Y Y

OMKTS Increase or maintain market share Y Y N Y Y Y Y N Y Y Y Y Y

OMKTCG Introduce products to new customer groups Y Y N Y Y Y Y N Y Y Y Y N

OMKTGM Introduce products to new geographic markets Y Y N Y Y Y Y N Y Y Y Y N

SGALA Graphic arts/ layout/ advertising Y N N Y Y Y N N N Y Y Y Y

SDOS Design of objects or services Y N N Y Y Y N N N Y Y Y Y

SMED Multimedia Y N N Y Y Y N N N Y Y Y Y

SWDS Web design Y N N Y Y Y N N N Y Y Y Y

SSWD Software development Y N N Y Y Y N N N Y Y Y Y

SMKR Market research Y N N Y Y Y N N N Y Y Y Y

SENAP Engineering / applied sciences Y N N Y Y Y N N N Y Y Y Y

SMSDM Mathematics / statistics / database management Y N N Y Y Y N N N Y Y Y Y

TURN08 Total turnover in 2008 Y Y N N N Y Y Y Y Y Y Y Y

TURN10 Total turnover in 2010 Y Y Y N N Y Y Y Y Y Y Y Y

EMP08 Total number of employees in 2008 Y Y N N N Y Y Y Y Y Y Y Y

EMP10 Total number of employees in 2010 Y Y Y N N Y Y Y Y Y Y Y Y

EMPUD % of employees in 2010 with a university degree Y N N N N Y Y Y Y Y N Y Y

Documents

Workstream (D) – Impact Analysis - European Commission · 1ESSnet on Linking of Microdata on ICT Usage, Grant number 50701.2010.001-2010.578 and ICT Impacts Grant number 49102.2005.017-2006.128