Micro Moments Database for cross‐country analysis of ICT, innovation, and economic outcomes

• Author: Michael Polder,Eric Bartelsman,Eva Hagsten
• DOI: http://doi.org/10.1111/jems.12256
• Date: 01 September 2018
• Published date: 01 September 2018

Received: 24 September 2015 Revised: 5 January 2017 Accepted: 5 January 2017

DOI: 10.1111/jems.12256

ORIGINAL ARTICLE

Micro Moments Database for cross-country analysis of ICT,

innovation, and economic outcomes

Eric Bartelsman1,2, 3 Eva Hagsten4Michael Polder5

1Department of Economics, VU University

Amsterdam, Amsterdam, The Netherlands

2TinbergenInstitute, Amsterdam, The

Netherlands

3IZA Bonn, Germany

4Universityof Iceland

5Statistics Netherlands

Correspondence

Eric Bartelsman, Department

ofEconomics, VU University

Amsterdam,Amsterdam, The Netherlands.

Email:ebar telsman@feweb.vu.nl

Abstract

This paper provides technical documentation to a database built up from firm-level

sources titled Micro Moments Database (MMD) that is made available forresearchers

through Eurostat. The MMD is an internationally harmonized research database of

statistical moments collected from linked longitudinal firm-level data in a large selec-

tion of EU national statistical offices. The underlying sources for the database are

business registers, firm-level surveys on production, usage of Information and Com-

munications Technologies (ICT) and innovative activities, as well as recorded infor-

mation on trade and worker education, all linked at the firm level. The unit of observa-

tion in the MMD represents groups of firms within industries and allows research that

bridges micro and macroanalysis. The paper delineates the type of research questions

that uniquely can be addressed with the MMD, and the advantages and disadvantages

of using MMD for questions where alternative data sets are available. The paper next

presents the methodology underlying construction of the MMD and provides docu-

mentation of the rich set of features. Finally, the paper provides descriptive statistics

that highlight the unique character of the data and reviews some of the cross-country

analytical work already conducted using the MMD.

KEYWORDS

cross country data, firm performance, innovation, ICT, productivity

1INTRODUCTION

There is a long and winding road leading from an innovation to its ultimate impact on society. Most of the twists and turns

are fertile ground for economic research, with open questions about incentives and impediments to innovative activity, and the

effect of competition and cooperation in going from adoption to private and social impact. The path is flanked by regulations,

policies, subsidies, and taxes, whose impact on innovation, intended or unintended, also are subject of much research. In order

to make progress in understanding the process of innovation, its impact on society, and the role played by policy and economic

environment, data are needed to illuminate the area of research. A particular area where data were lacking, was in tracking

how firm-level decisions regarding innovative activities, through interactions with customers, suppliers, and competitors in the

This work has been made possible by Eurostat Grant Agreement 50721.2013.001-2013.082. The Micro Moments Database is available at Eurostat

(http://ec.europa.eu/eurostat/web/microdata/micro-moments-dataset). We thank participants and a discussant at the Searle Center 2015 for their valuable com-

ments and suggestions.

626 © 2018 Wiley Periodicals, Inc. J Econ Manage Strat. 2018;27:626–648.wileyonlinelibrary.com/journal/jems

BARTELSMAN ET AL.627

marketplace, lead to aggregate outcomes. For example, analytical workwit h the Micro Moments Database (MMD) can be used

to disentangle interactions between firm-level use of information and communication technology (ICT) or firm-level innovative

activities on the one hand and aggregate productivity or aggregate employment on the other.

The MMD is designed to loosen the data constraint for such research. The MMD contains some novel indicators on ICT-use

and innovative business practices. More importantly, the MMD provides a facility to identify how changes in national level

policies or national economic environment affect the road from innovation to impact. Especially, the ability to track both behav-

ioral characteristics and aggregate outcomes in a cross-country setting, with variation in policy stance as well as in timing and

direction of policy changes, provides an opportunity for policy effects to be identified. Owing to legal constraints on bringing

together single country data sets containing firm-level linked registers and surveys, the MMD makes the trade-off of gaining a

cross-country dimension for analysis at the expense of aggregating up from the firm-level to within-industry groups of repre-

sentative firms.

The MMD is sourced from firm-level data available at national statistical agencies in European countries. The underlying

firm-sources are business registers (BRs) with the universe of firms, firm-level surveys on production, innovation, ICT-use as

well as recorded information on trade and worker skills that are linkable to firms. There are approximately20 million fir ms in the

underlying BRs, and on the order of 1/2 million firms when these are linked to the firm-level surveys on ICT-use and innovation.

The value added covered by firms used for the MMD amounts to well over 50% of total industry value added in these countries

and the countries included cover more than 80% of valued added of the EU-28 plus Norway. In an internationally harmonized

manner, the firm-year observations from the national sources are aggregated such that results can be pooled into the cross-country

MMD data set.

The MMD can be linked at the industry-level to other international industry-level data sets, such as STAN or EUKLEMS, yet

provides information below the industry-level that can be used to identify producer behavior. Also, the MMD can be merged at

the record level to custom aggregations from the firm-level, for exampleby a researcher with access to confidential U.S. Census

business registers and surveys. The MMD includes industry-level aggregates typically published bystatistical agencies, such as

sums of employment or value added and means of productivity or capital intensity, but also higher moments of the firm-level

distributions of underlying variables. Uniquely, the MMD contains moments from multivariate distributions, such as output or

growth of firms for different bins of firm-age, size, export intensity, or innovationprofiles. Fur thermore, by combining variables

from different surveys at the firm level, the MMD contains novelindicators, such as industry average ICT intensity or percentage

of ICT-schooled workers in an industry, that go beyond what is available or can be derived from the survey data alone in each

country.

The MMD emerged from three subsequent Eurostat-funded projects on microdata linking. In these projects, researchers and

statisticians from 14 European National Statistics Offices (NSOs), worked together with academic partners to clean and harmo-

nize the underlying sources.1The projects made use of a protocol called distributed microdata analysis (DMD, see Bartelsman,

Haltiwanger, & Scarpetta, 2004) to execute common computer code in each country to generate the MMD. The MMD data set

has been made available for researchers through Eurostat (see Table B1).

The paper starts by delineating the types of research questions that uniquely can be addressed with the MMD, and the pros

and cons of using MMD for research questions where alternate data sources are available. Next, the paper provides a detailed

methodology of the route from national firm-level sources to construction of the MMD. The following section provides docu-

mentation of the rich set of features of the MMD, and information on how to get access to the data. Finally, the paper provides

descriptive statistics that highlight the unique character of the data and reviews some of the cross-country analytical workalready

conducted using the MMD. These results are novel in the sense that no cross-country evidence of this type has been available

previously. The examples are chosen not only to provide evidence on current academic and policy questions relating to ICT

and innovation, but also to showcase the types of analysis that presently can be done using the MMD. Especially in the area

of assessing the aggregate impact of the policy environment, or evaluating the effects of policy changes, having cross-country

indicators of both firm-level responses and market outcomes is crucial for identification.

2DATA FOR RESEARCH ON ICT, INNOVATION

The MMD includes information on traditional firm-level innovative activity, such as R&D spending or product and process

innovation as captured by the Community Innovation Survey (IS).2The MMD further combines information on innovative

use of ICT and ICT-enabled business processes at the firm level as captured by the EU ICT Usage in Enterprises Survey.3

The MMD also has information on nominal and deflated productive inputs—capital, (skilled) labor, materials—and outputs

and information on firm characteristics such as industry, ownership, and age. These are the types of information typically