University of Toronto Economics Case Study


1. what are the economic questions addressed by the paper. (explain each question and the key term.)
2. How do the authors motivate their paper? Why was the paper written?
3.How is the paper related to the literature?(provide reference of the literature and how is this paper relate to it)
Data used in the econometric analysis
1.a. If observational data, tell us what kind (cross-section, panel or time series). What years are covered in the data set and who is observed (individuals, households, firms, etc)
b. If experimental data, what kind of experiment (social, laboratory or field experiment)

1 attachmentsSlide 1 of 1attachment_1attachment_1.slider-slide > img { width: 100%; display: block; }
.slider-slide > img:focus { margin: auto; }

Unformatted Attachment Preview

American Economic Review 2020, 110(4): 1055–1103
Industrial Espionage and Productivity†
By Albrecht Glitz and Erik Meyersson*
In this paper, we investigate the economic returns to industrial
espionage. We show that the flow of information provided by East
German informants in the West over the period ­1970–1989 led to
a significant narrowing of sectoral TFP gaps between West and
East Germany. These economic returns were primarily driven by
relatively few ­high-quality pieces of information and particularly
large in sectors closer to the West German technological frontier.
Our findings suggest that the E
­ ast-to-West German TFP ratio
would have been 13.3 percent lower at the end of the Cold War
had East Germany not engaged in industrial espionage in the West.
(JEL L16, N44, O33, O38, O47, P24)
Despite the rich history of illicit technology transfer and its significant contemporary importance, industrial espionage and its associated costs and benefits have
received little attention in the economic literature.1 Undoubtedly, the secret nature
* Glitz: Universitat Pompeu Fabra, IPEG, and Barcelona GSE (email:; Meyersson:
Handelsbanken Capital Markets (email: Stefano DellaVigna was the coeditor for this article. We would like to thank representatives of the Agency of the Federal Commissioner for the Stasi Records
(BStU) and Gerhard Heske for sharing the data and providing valuable expertise throughout this project. We are
grateful to Niklas Flamang, Blanca Gil Rosell, Stefán Gudmundsson, Adrian Lerche, Chris Schroeder, and Paul
Soto for excellent research assistance, and Andreas Ziehe for help with the machine learning part of the paper. We
thank three anonymous referees for very helpful comments and suggestions. We are grateful to Philippe Aghion,
Eli Berman, David Card, Ruben Durante, Gabrielle Fack, Martin Feldstein, Tarek Ghani, Rudolph Glitz, Mariko
Klasing, Gianmarco Leon, Giacomo Ponzetto, Uta Schönberg, Alessandro Tarozzi, Manuel Trajtenberg, John Van
Reenen, Felix Weinhardt, and numerous seminar participants for valuable feedback on earlier drafts of the paper.
Albrecht Glitz gratefully acknowledges financial support from the Barcelona GSE (through the Program “Severo
Ochoa” for Centers of Excellence in R&D, SEV-2015-00563, funded by the Spanish Ministry for Economy and
Competitiveness) and the Ministry of Science, Innovation and Universities (through the National Programme
for the Promotion of Talent and Its Employability, the Ramón y Cajal grants, MINECO-RYC-2015-18806). He
also thanks the German Research Foundation (DFG) for funding his Heisenberg Fellowship (GL 811/1-1) and
Alexandra Spitz-Oener for hosting him at Humboldt University Berlin. The views, analysis, conclusions, and any
remaining errors in this paper are solely the responsibility of the authors.
Go to to visit the article page for additional materials and author
disclosure statements.
While “industrial espionage” and “economic espionage” are often used interchangeably, some authors draw
a distinction between them with industrial espionage referring specifically to activities conducted by individual
companies against their competitors for commercial purposes and economic espionage referring to activities in the
economic domain conducted on behalf of foreign governments and for reasons that are not exclusively commercial.
Because of the distinct focus on different industry sectors in our analysis, we have followed the common practice in
the context of East German ­scientific-technical espionage of using the term “industrial espionage” throughout the
paper (see Müller, Süß, and Vogel 2009). Note that from a legal point of view, there is no uniform definition of what
constitutes the punishable offense of espionage. In the United States, the Economic Espionage Act of 1996 defines
economic espionage as “the theft or misappropriation of a trade secret with the intent or knowledge that the offense
will benefit any foreign government, foreign instrumentality, or foreign agent.” In Germany, espionage is typically
APRIL 2020
of the practice obscures its economic significance which is believed to be considerable. For example, industrial espionage is currently estimated to cost the US economy around $19 billion per year and the German economy around 11.8 billion euros
per year, both figures from the lower end of a wide range of available estimates.2
Compared to the costs, the economic benefits accruing to those countries actively
engaging in industrial espionage are even more opaque. However, its persistent and
widespread use as a channel for technology transfer suggests that these benefits are
In this paper, we provide the first comprehensive analysis of the relationship
between ­state-sponsored industrial espionage and technological progress. The historical setting is the Cold War period in which industrial espionage became instrumental for economic development as the communist bloc attempted to catch up
with the capitalist world’s technological advantage. The centerpiece of our study
is a dataset, the ­so-called SIRA, that comprises the entire stock of information
East German foreign intelligence sources gathered abroad during the period 1970
to 1989. This unique database includes detailed information on 189,725 individual pieces of information received by the East German Ministry for State Security
(MfS, commonly referred to as the Stasi), including their precise date of receipt, the
code names of their sources, and a list of keywords describing each item’s content.
To operationalize this wealth of data, we use the keywords provided to attribute each
piece of information to the appropriate industry sector(s). We then merge the aggregated ­sector-specific information flows to sectoral total factor productivity (TFP)
measures which we compute from time series data on sectoral gross value added,
employment, and gross fixed capital investment. In our main estimation equation,
we regress changes in sectoral log TFP gaps between West and East Germany
(equivalent to differences in TFP growth rates) on past inflows of s­ector-specific
information generated by industrial espionage, controlling for direct measures of
R&D activity in both parts of Germany and their initial distance to the technological
frontier. Our estimates thus speak directly to the question in how far industrial espionage allowed the East German economy to keep up with technological progress in
the West.
Our results provide evidence of significant economic returns to industrial espionage, indicating an important role of international knowledge flows for productivity growth in laggard countries. A 1 standard deviation increase in the inflow of
information results in a 7.3 percentage point (4.9 percent) decrease of the log TFP
gap and a 5.5 percentage point (4.3 percent) decrease in the log output per worker
gap between West and East Germany. We also provide complementary evidence
for a positive effect of industrial espionage on a purely q­ uantity-based measure of
output and on the number of new goods produced in East Germany. Furthermore,
we show that industrial espionage tended to crowd out investments in regular overt
R&D in East Germany. To address potential endogeneity concerns, we employ two
distinct instrumental variable strategies. The first one utilizes information generated
by informants who were already active at the beginning of the sample period in a
punishable under §99 StGB according to which anyone is subject to prosecution who “discloses or delivers facts,
objects or findings to a foreign intelligence service, or agrees to such activity.”
Sources: Munsey (2013) for the United States and Corporate Trust (2014) for Germany.
VOL. 110 NO. 4
s­hift-share-type setting. The second exploits the sudden disappearance of certain
informants as providers of information as an exogenous source of variation. Both
instruments lead to results that are somewhat larger in magnitude than our baseline
ordinary least squares (OLS) estimates.
In a series of robustness checks, we show that our findings are not exclusively
driven by the very prominent IT sector and robust to variations in the way observations are weighted and pieces of information assigned to different sectors. We also
provide evidence that changes in the calibration of the key parameters underlying
our sectoral TFP measures have little impact on our estimates. Through a series
of placebo tests, we further demonstrate that our main results are not driven by a
spurious correlation between the regressor of interest and the dependent variable,
a possibility given their specific functional forms. When testing the robustness of
our findings to alternative functional forms, for example by changing the applied
normalization or using raw instead of scaled measures of espionage inflows, the
results are more mixed. While generally maintaining the expected sign, the estimates become statistically insignificant on a number of occasions, especially in the
IV estimations. Contrary to our baseline model, these specifications have no theoretical foundation, but this lack of significance does indicate some sensitivity of our
findings to alternative functional form assumptions.
Analyzing different dimensions of heterogeneity, we document that the positive
effect on East German productivity growth is primarily driven by relatively few
­high-quality pieces of information and that industrial espionage was particularly
effective in those sectors that were closest to the West German technological frontier. We conclude by running a counterfactual simulation of how East German TFP
would have evolved in the absence of industrial espionage, showing that it had overall a noticeable but quantitatively modest mitigating effect on the productivity gap
with West Germany. Our findings suggest that the ratio of ­East-to-West German
TFP, which amounted to 21.8 percent in 1989, would have been 13.3 percent lower
(so 18.9 percent) in the absence of industrial espionage. For some sectors, however,
we find that industrial espionage was vital to avoid a significant further opening of
the technological gap. In the electronics sector, for example, the already low East
German TFP level relative to West Germany’s of 12.0 percent in 1989 would have
been 39.2 percent lower (7.3 percent) if East Germany had not been so prolific in
acquiring relevant technological information in this sector through its espionage
activities in the West. A tentative ­cost-benefit analysis indicates that the net return of
industrial espionage was substantial, with annual benefits of the order of 10.1 billion
euros contrasting with annual running costs of around 11.0 million euros.
Besides providing the first empirical assessment of the role of industrial espionage
for technological progress, our paper speaks to several existing literatures in economics. Most importantly, since industrial espionage inherently involves the flow of
technological knowledge from the targeted to the perpetrating country, our findings
contribute to the extensive work on international technology diffusion.3 This literature
has focused either directly on international R&D spillovers (e.g., Jaffe, Trajtenberg,
and Henderson 1993; Griffith, Harrison, and Van Reenen 2006; Coe, Helpman, and
For overviews of this literature, see Jones (2005), Keller (2010), and Comin and Mestieri (2014).
APRIL 2020
Hoffmaister 2009) or studied the role of international trade (e.g., Eaton and Kortum
2002; Cameron, Proudman, and Redding 2005; Buera and Oberfield 2016), foreign direct investment (e.g., Javorcik 2004; Keller and Yeaple 2009; Guadalupe,
Kuzmina, and Thomas 2012), and international migration (e.g., Hornung 2014;
Moser, Voena, and Waldinger 2014) as possible conduits of knowledge spillovers.
Contrary to most of this work, we observe knowledge flows directly which allows
a more accurate assessment of their importance for productivity growth. Due to
the different markets in which East and West Germany operated at the time, the
spillover effects we estimate are also less likely to be confounded by countervailing business-stealing effects through product market rivalry, a lingering identification problem in past studies on the impact of R&D spillovers on economic growth
(Bloom, Schankermann, and Van Reenen 2013; ­Fons-Rosen et al. 2017).
Viewing industrial espionage as a means of acquiring new s­cientific-technical
knowledge, our study also relates to the literature on the role of innovation in
explaining productivity growth (e.g., Aghion and Howitt 1992; Hall, Mairesse,
and Mohnen 2010). In analyzing the heterogeneous effects of industrial espionage
across East German industries and its impact on East Germany’s own R&D efforts,
we also touch on the literatures on absorptive capacity (e.g., Aghion and Jaravel
2015) and the role of the distance to the technological frontier for aggregate productivity growth, technology adoption, and innovation (e.g., Griffith, Redding, and
Van Reenen 2004; Acemoglu, Aghion, and Zilibotti 2006; Comin and Hobijn 2010).
Apart from the broader innovation literature, our analysis also contributes to the
literature studying the social and economic consequences of covert activities and
secrecy. In a recent study, Lichter, Löffler, and Siegloch (2016) exploits discontinuities at state borders within East Germany to show that higher levels of Stasi surveillance during the 1980s led to lower levels of social capital and worse economic
outcomes in the p­ ost-unification period. Other examples for the adverse effects of
secrecy come from the archival study of the former Soviet Union’s intelligence
agency, the KGB, revealing for instance that secrecy incurs broad efficiency costs
in the economy (Harrison 2008). In the US context, declassified intelligence documents have been used to show that C
­ IA-supported coups led to significant stock
market gains for firms with a particular interest in regime change (Dube, Kaplan,
and Naidu 2011) and that imports from the United States increased systematically
in those countries in which the CIA successfully helped install a new leadership
(Berger et al. 2013). Finally, in studying the effects of an arguably widespread but
generally unobservable economic activity, our paper also has some connection to the
literature on the shadow economy, which has provided insights into similarly elusive activities such as tax evasion (e.g., Kleven et al. 2011), corruption (e.g., Olken
2007), and illicit trade (e.g., Fisman and Wei 2009).
Outside of economics, there is of course a more extensive literature on espionage
by historians, often focusing on specific case studies or the successes and failures
of individual spies (e.g., Friis, Macrakis, and ­Müller-Enbergs 2009). Regarding
East German espionage in the West, Herbstritt (2007) provides a comprehensive
picture of the recruitment strategies of the Stasi and the social structure of its network of informants, complementing the extensive work on the Stasi and its foreign
intelligence branch by ­Müller-Enbergs (1996, 1998, 2011). Macrakis (2008) comes
closest to the type of question we analyze in this paper, arguing that the Stasi’s
VOL. 110 NO. 4
s­ cientific-technical intelligence activities were ultimately a failure as the secretive
nature of ­high-tech espionage clashed with the openness required for successful
scientific development. Yet as late as 1989, East Germany was seen by some as
“communism that works” and “the communist world’s ­high-technology leader… its
capital goods known for quality workmanship.” 4 Our main results show that once
the entirety of the information flows from the West are taken into account, East
Germany’s industrial espionage program can by all means be viewed as a success.
The rest of the paper is organized as follows. Section I provides the historical context in which East Germany engaged in industrial espionage in the West.
Section II describes the various data sources used in the paper. Section III presents two case studies that illustrate the process through which industrial espionage
affected production in East Germany. Section IV introduces the empirical framework and estimation strategy. Section V presents the main results as well as further
complementary analysis. Section VI concludes the paper.
I. Historical Background
East German industrial espionage was to a large extent a response to the West’s
implementation of economic containment policies at the onset of the Cold War.
Already shortly after the end of World War II, Western Bloc countries led by the
United States imposed a trade embargo on their Eastern Bloc counterparts, initially focusing on restricting the trade of arms and weapons technology. Over the
following decades, the Coordinating Committee for Multilateral Export Controls
(CoCom) served as a tool for the West to implement ever more stringent export controls on goods bound for the communist East. Increasingly, these included not just
goods from the military and nuclear sectors but also industrial “­dual-use” products
that could, at least in principle, be used for military purposes. As the trade embargo
against the communist bloc intensified, East Germany came to rely increasingly on
its industrial espionage to keep up with the West.
The Stasi’s industrial espionage was conducted predominantly under its foreign
intelligence unit (Hauptverwaltung Aufklärung, HVA), led by the famous spy chief
Markus Wolf. The branch in charge of gathering ­scientific-technical information
in the West was the Sector for Science and Technology (Sektor Wissenschaft und
Technik, SWT), which by the end of 1988 comprised around 260 ­full-time staff
members and consisted of three specialized departments responsible for the acquisition of information in the areas of Energy, Biology and Chemistry (Abteilung XIII),
Electronics and Electrical Engineering (Abteilung XIV ), and Machine Building and
Embargo Goods (Abteilung XV ), one department responsible for the evaluation of
all incoming information (Abteilung V ), and a number of smaller working groups
(­Müller-Enbergs 1998).
For the collection of ­scientific-technical information, the Stasi relied on an extensive network of informants in Western Bloc countries, especially West Germany.
Knabe (1999), ­Müller-Enbergs (1998), and Herbstritt (2007) provide insightful information about the recruitment, motivation, and social background of the
“East Germany Losing Its Edge,” New York Times, May 15, 1989,
APRIL 2020
Stasi’s collaborators in West Germany. More than half of the informants still active
at the end of the 1980s were initially brought to the Stasi’s attention as potential
recruitment targets by other already active informants. In contrast, informants who
approached the Stasi on their own initiative constituted only a relatively minor fraction of less than 5 percent. Internal Stasi documents further show that 60 percent
of the informants were recruited primarily due to their p­ olitical-ideological convictions, 27 percent due to material interests and only less than 1 percent “under pressure.” According to Herbstritt (2007), a simple informant in the West could expect
monthly payments of between 100 and 500 Deutsche Mark plus the reimbursement
of expenses, which would represent a moderate ­top-up of their regular salaries. For
high profile informants, however, these regular payments could be substantially
higher, reaching amounts of several thousand Deutsche Mark per month. In terms of
­socioeconomic background, most of the informants involved in industrial espionage
in the West were ­middle-aged male salaried employees, predominantly engineers
or employees with science degrees, although a number of sources also worked in
personnel departments or as businessmen. These informants were not necessarily
leaders in their field or heads of departments but often more m
­ id-ranking employees like engineer Dieter Feuerstein (codename: Petermann) at MBB, who passed
on ­top-secret military plans, Peter Alwardt (codename: Alfred), who worked as an
engineer at AEG/Telefunken, and Peter Köhler (codename: Schulze), who worked
for Texas Instruments.
In terms of s­ cientific-technical fields targeted, the Stasi generally cast a wide net.
Of broad interest were, for example, processes for a more economical use of energy
or more efficient processing techniques for raw materials (­Müller-Enbergs 1998).
A particularly important role in the Stasi’s industrial espionage program, however,
was given to the electronics sector, especially since the 1970s when the East German
political leadership decided to become a world leader in computer technology and
started to direct significant resources to the production of microchips and the infiltration of Western electronics companies such as IBM and Siemens. Meanwhile,
Western intelligence in East Germany remained by most accounts limited, especially in the economic sector which was technologically behind and therefore not
a priority target of Western espionage. As such, the transfer of technologies was
overwhelmingly a ­one-way street.
II. Data
A. SIRA Data
Our main data source on the Stasi’s industrial espionage activities in the West
is the HVA’s central electronic database SIRA (System der Informationsrecherche
der Hauptverwaltung Aufklärung), currently maintained by the Agency of the
Federal Commissioner for the Stasi Records (BStU). Within SIRA, subdatabase 11
(Teildatenbank 11) comprises records of essentially all ­scientific-technical information that the Stasi’s informants passed on to the HVA during the 1970s and 1980s.5
In anticipation of the introduction of SIRA, the HVA started in 1968/1969 to systematically record all
incoming information on punched tape, which was then fed into the SIRA database when it was launched in July
VOL. 110 NO. 4
Given the historical circumstances, the fact that these data still exist is remarkable.
At the beginning of 1990, with political changes sweeping through East Germany,
it was decided to disband the Stasi and physically destroy all sensitive information,
including all electronic data carriers. By March 19, 1990, “10,611 magnetic tapes,
5,267 disks, 544 removable hard disks, and 80 sacks of loose magnet tape material” had been destroyed, including all data stored in the original SIRA system.6
However, in the process of a comprehensive data conversion of the entire SIRA system in 1988/1989, the HVA had made copies of the original data which were then
overlooked when the Stasi was liquidated. The data from these copies, meticulously
reconstructed by the BStU during the 1990s (see Konopatzky 2007), form the basis
of the present analysis.
In total, 189,725 pieces of information were recorded in SIRA between 1968 and
1989, corresponding to an annual average inflow of 8,624 items. Online Appendix
Figure A1 displays the distribution of this flow of information over time. Throughout
the 1970s, the volume of information received per year was declining but started to
increase steadily again from 1979 onward, eventually peaking in 1988, the year
before the fall of the Berlin Wall and the last year fully covered by SIRA, with a
record of 15,658 pieces of information.7 Given these magnitudes, it is not surprising
that not all of the information flow from the West necessarily involved the theft or
misappropriation of trade secrets, the legal definition of economic espionage in,
for example, the United States today. Instead, a significant fraction of the intelligence received likely referred to information that was publicly available anyway,
thus more resembling s­ o-called “competitive intelligence” which is generally considered to be legal. While we do not have a direct way of discerning whether a
given piece of information was obtained illegally, internal quality assessments of the
Stasi provide some indication about the intrinsic value of each piece of information
(see Section VE) which should also reflect the difficulty of obtaining the relevant
Upon arrival at the Stasi, specialist internal evaluators created, for each incoming piece of information, an electronic entry in the SIRA database in which they
recorded, among other things, the date of arrival of the information, the source of
the information, as well as a number of often highly specific keywords to describe
the information’s content. After this initial documentation, the received material was
then passed on to potentially interested parties, typically s­tate-run enterprises or
East German research facilities, for further assessment and economic exploitation.
Contrary to the electronic data entries in SIRA, the original intelligence delivered
(documents, photos, tapes, disks, blueprints, etc.) was destroyed in the process of
disbanding the Stasi in early 1990, so that only the data entered into the SIRA system can shed light on the actual content of each piece of information received. In
1974. Industrial espionage on behalf of the Stasi in the West was of course already taking place prior to 1968 but
there are no electronic records that would allow us to extend our analysis to this earlier period.
Source: Komitee zur Auflösung des AfNS: Abschlussbericht der Vernichtung der magnetischen Datenträger zu
personengebundenen ­EDV-Projekten des ehemaligen AfNS, vom 19.03.1990, BArch, DO 104.
While the SIRA data do not allow determining the country of origin of a given piece of information, internal
documents of the Stasi as well as other historical sources show that West Germany was by far the most important target of the Stasi’s espionage activities. According to ­Müller-Enbergs (2011), 82.7 percent of the informants
abroad that were handled by the three principle departments of the HVA’s Sector for Science and Technology in
December 1988 were located in West Germany.
APRIL 2020
total, subdatabase 11 comprises 143,005 distinct keywords, 68.5 percent of which
are only used once over the entire time period. On average, each piece of information is described by 5.6 distinct keywords but the distribution is skewed to the right,
with a median of 5, a ­ninety-fifth percentile of 10, and a maximum of 145 keywords.
To operationalize these keywords and connect them to the sectoral time series
data, we selected in a first step the 2,000 most frequently occurring keywords,
which together account for 63.8 percent of all keyword entries in the database, and
assigned them to their corresponding sectors. Online Appendix Table A1 lists the 30
most frequently and 10 least frequently used keywords in this subsample, together
with their English translations, their frequency in the data, and the sectors to which
we allocated them. Examples of frequently used keywords are Military Technology,
Electronics, Chemistry, Microcomputer, Metallurgy, Optics, IBM, and Nuclear
Power Plant. Overall, we were able to assign 55 percent of the 2,000 most common
keywords to at least one of the 16 sectors for which we have information on output, employment, and investment.8 After this allocation procedure, the vast majority of the distinct pieces of information in our sample are described by between
1 and 5 ­sector-specific keywords, and only 18.6 percent are not described by any
­sector-specific keyword. Online Appendix Table A2 provides a number of concrete
examples that illustrate the allocation procedure.
Figure 1 shows the sectoral distribution of the 151,627 pieces of information that
could be allocated to at least one of the 16 available sectors over the period 1968 to
1989. In our baseline specification, we count a piece of information as pertaining
to a specific sector if it is described by at least one keyword corresponding to that
sector. A given information may therefore refer to more than one sector. In line
with historical accounts, the sector Office Appliances, Computers, and Electronics
constituted by far the most important sector for industrial espionage, with 100,279
pieces of related information in total, followed by the sectors Chemicals (33,409),
Utilities (23,485), and Machine Building (23,152). For our empirical analysis, we
drop the early years 1968 and 1969 as well as the final year 1989, since these are
only partially covered by the SIRA data.
Looking at the providers of these pieces of information, the SIRA database
identifies 2,968 distinct informants based on their assigned registration numbers.
Online Appendix Table A3 lists the 20 most productive sources of information over
the period 1968 to 1989. However, these ­top-ranking informants were certainly an
exceptional group in terms of the amount of information they generated. Across the
whole group of informants, the median and mean inflow of information amounts to
only 4 and 52.3 items respectively, reflecting the highly ­right-skewed distribution
illustrated in online Appendix Figure A2. The information provided by most informants throughout their time in the service of the Stasi was thus limited, reflecting
the cautious approach by the Stasi in handling its sources as well as the difficulties
for most informants to tap into relevant information. Online Appendix Figure A3
depicts the distribution of the first and last active year in which each informant is
observed in the data. The left panel suggests that recruitment of new informants was
The remaining 45 percent are either not classifiable (80.9 percent) or refer to other sectors of the economy such
as agriculture, construction, automobile repairs and consumer goods, transportation and communication, finance,
leasing and public and private services, health, military, or the aerospace industry (19.1 percent).
VOL. 110 NO. 4
Food and tobacco
Textiles and clothing
Leather products
Paper, printing, and publishing
Furniture, jewelry, and music instruments
Coking and petroleum
Rubber and plastics
Glass, ceramics, and non?metallic minerals
Machine building
Office appliances, computers, and electronics
Motor vehicles
Utilities: energy and water supply
Frequency (1,000s)
Figure 1. Sectoral Distribution of Information
Note: Figure shows the ­sector-specific inflows of information received by the HVA between 1968 and 1989.
an ongoing process, with increasing efforts from the ­early 1980s onward. The right
panel shows that informants also continuously ceased to provide further information. We will exploit this fact later on in the construction of one of our instrumental
B. ­Industry-Level Data
The second key data source for our empirical analysis are the ­sector-specific time
series for gross value added, total employment, and gross fixed capital investment
constructed by Heske (2009, 2013, 2014). The purpose of this work was to provide
a comparable, retrospective accounting of the development of key economic indicators for different industry sectors in West and East Germany over the time period
1950 to 2000. Due to the fundamental differences in economic systems before
German unification in 1990, such computations constitute a challenging task, not
least because West and East Germany followed different national accounting standards during the ­pre-unification period.9
While West Germany’s national accounting was based on the nowadays-standard System of National Accounts
(SNA), East Germany