The Empirical Basis for Statutory Database Protection
After the European Database Directive
by Matt Block
The debate: whether to increase protection for
databases in the United States.
In the debate over whether the U.S. should enact database protections similar to
those in Europe, the rhetoric and philosophical arguments have been prominent and the
facts themselves downplayed. A survey of the available evidence in 2002, for instance,
noted that there was no good evidence yet adduced of many of the empirical factors in the
debate.1 This is as ironic as it is understandable: it is the facts themselves that would
receive protection under database legislation, but the facts themselves are difficult to find
and track. Nonetheless, the facts are critical. Some scholars have noted that any policy
change may create substantial unintended consequences. Thus, compelling empirical
evidence should be required before dramatic policy changes are undertaken.2
Instead of such evidence, both proponents and opponents have argued using
compelling economic theory. Proponents have argued that measures that benefit database
producers also benefit the public at large; there is no conflict between the interests of the
two groups.3 This is because an optimal output decision on the part of database
producers will also be efficient. It will create sufficient incentives to produce new
databases, but will allow consumers to purchase databases at a reasonable rate. However,
they have argued, databases have characteristics of public goods, and must be treated
differently as a result. In particular, databases have a declining marginal cost, because
Justin Hughes, “Political Economies of Harmonization: Database Protection and Information Patents,”
Cardozo Law School: Jacob Burns Institute for Advanced Legal Studies, Research Paper Series No. 47
(August 2002). Available at http://ssrn.com/abstract_id=318486.
2
See Hughes, supra note 1.
3
See Braunstein, infra.
1
1
the cost of reproduction or additional access is approximately zero. Thus, database
producers need some mechanism to protect and recoup their initial investment in database
production against second-comer free-riders.
Opponents argue that the free-rider problem is overstated and easily solved with
self-help and market mechanisms and existing legislative protection.4 They argue that
databases are inputs into other productive activity, including further database production,
so that any legislative remedy to the “free-rider” problem may ultimately lead to
decreased database production and quality, and to lower productivity in other areas.
Indeed, some opponents argue even more forcefully: because market economies depend
on the free flow of information, increased protection may actually bind the invisible hand
of market forces, and completely upset the existing incentive system.5 Thus, protectionist
measures may not even have the limited incentive value the proponents suggest, and they
will certainly increase the costs of developing (and therefore decrease the production of)
derivative databases, which use existing databases as inputs.
In addition to the economic arguments, commentators on both sides make moral
and natural rights arguments for their positions. Proponents argue forcefully that anyone
who invests in the creation of a thing has a natural right to recoup that investment. They
argue that no one would ever advocate allowing purveyors of physical goods to simply
repackage their competitor’s works, and that allowing “pirates” to “free-ride” on their
creative efforts amounts to the same thing. Opponents argue that information is different
See generally Stephen M. Maurer, “Raw Knowledge: Protecting Technical Databases for Science &
Industry,” A Report for the National Research Council’s Workshop on Promoting Access to Scientific and
Technical Data for the Public Interest, (January 1999).
5
See generally James Boyle, Chapter 4: Information Economics, Shamans, Software, and Spleens: Law
and the Construction of the Information Society, pp. 35-46 (1996).
4
2
from property, that the normally cognized rights in information are entirely statutory and
not natural, and that information “wants to be free”.6
Where is the evidence?
As it happens, little empirical evidence has been produced on either side of the
debate. Proponents of database protection argue that there is not enough database
production today, and that increased protection will create incentives for new creation.7
Because there are other proposals, however, they also argue that increased protection is
the best way to create these incentives.8 They argue that increased database production
will be good for society.9 However, proponents have offered no evidence that these
things are actually happening, or will actually happen.
On the other side, opponents of increased protection argue that increased
protection will disincent creators to compete on quality, create awkward new barriers to
entry, and raise the costs of creation for second-comers.10 On the net, they argue, this
may actually lead to less, and lower quality, database creation. Opponents too, however,
have been silent as to significant examples of their dire predictions coming to pass. The
reason for the silence of both opponents and proponents is the difficulty in obtaining the
facts.
For an overview of the arguments and other insights, see J.H. Reichman, Pamela Samuelson, “Intellectual
Property Rights in Data?,” Vanderbilt Law Review Vol. 50, p. 51 (January 1997).
7
See Yale M. Braunstein, “Economic Impact of Database Protection in Developing Countries and
Countries in Transition,” WIPO Standing Committee on Copyright and Related Rights, Geneva (April 4,
2002).
8
Id.
9
Id.
10
The critical difference may stem from opponents’ recognition of databases as an input to database
production. Compare Jonathan Band, Testimony Before the Subcommittee on Courts and Intellectual
Property of the United States House of Representatives Committee on the Judiciary on the Collections of
Information Antipiracy Act, H.R. 2652 (February 12, 1998) with Laura D’Andrea Tyson, Edward Sherry,
Statutory Protection of Databases: Economic and Public Policy Issues, available at
http://www.house.gov/judiciary/41118.htm.
6
3
As difficult as it may be to find and track, the evidence is available. Since 1998,
Europe has been undertaking a natural experiment in incentives to create. By 2001, all of
the EU countries had implemented, at least partially, the European Database Directive.11
The goal of the Directive was to harmonize protection for commercial databases, so that
they could be made available throughout the EU, and thereby generate higher profits that
would lead to greater incentives to create new databases. Because the databases sought
to be protected were commercial in nature, they are likely to be advertised and listed in
commercial directories.
The Gale Directory of Online, Portable, and Internet Databases is published each
year and tracks databases throughout the world that are made available publicly to
commercial users to view and access electronically.12 Admissible databases are either
available as a CD-ROM or on other media, or they are actually published online. There
is an important caveat; the Directory lists only commercial databases, and is silent as to
those made freely available.13 Thus, it lists a sample of the databases that are
commercially available, easily susceptible to automated copying, and subject to
protection under the American proposals and the European laws.
11
Different entities date the various implementations at different times, perhaps because implementations
themselves have been somewhat piecemeal in their coverage. The NautaDutilh Final Report sets the final
implementation date as being that of Luxembourg, which implemented the directive in 2001.
12
Gale Directory of Online, Portable, and Internet Databases. The Directory is part of the Gale Directory
of Databases, published on paper in two volumes each year. It is also available online via Westlaw,
http://www.westlaw.com/. It is not clear what the relationship between the paper Directory and the
electronic version have to one another, although the first volume of the paper Directory has the same title,
and perhaps the same contents, as the version available online. Due to limited access to the paper
Directory, no effort was made to verify this.
13
For discussion of the quantity of undercounting, see Clemente Forero-Pineda, “Scientific Research,
Information Flows, and the Impact of Database Protection on Developing Countries,” from Open Access
and the Public Domain in Digital Data and Information for Science, Proceeding of an International
Symposium, National Research Council of the National Academies (2004).
4
Is there a problem that new rights in databases will solve?
As good a resource as it is, the Directory cannot answer the preliminary policy
question, at least not alone. Before examining methods of remedying the database
underproduction problem one must first determine that there is, in fact, a problem and the
Directory cannot provide that negative evidence directly. It is worth noting, however,
that the United States produces and sells more databases than any other nation, and that
the growth of this industry is and has been greater than in Europe. If the United States is
facing the sort of database underproduction that might warrant legislative intervention
and the creation of new rights (and thus new burdens), it is certainly not the worst off of
the nations. Moreover, the database industry is economically viable in the United States,
and has experienced robust growth in the number of databases sold, and the revenue
generated thereby. For instance, the 2002 Economic Census indicated that over the 5year period from 1997 to 2002, the number of companies classifying themselves as
database publishers grew by around 25% (from 1,458 to 2,098), receipts grew by 50%
(from $12.3 B to $18.8 B), and payroll in the industry nearly doubled (from $1.7 B to
$3.2 B).14
In fact, there is little reason to believe that there is an essential market failure in
the United States leading to database underproduction. Database production has
traditionally been strong in this country, and it is getting stronger. For instance, in
response to a 1998 effort to create new database protections, one witness testified that the
number of databases in this country had increased 35% over a six year period, the number
of files within databases had grown 180%, and the number of online searches had grown
14
2002 Economic Census Information Industry Series, Directory and Mailing List Publishers, Report No.
EC02-511-04, November 2004. Available at http://www.census.gov/epcd/ec97/industry/E51114.HTM.
5
80% in the same time.15 Apparently US database providers have not been sufficiently
discouraged by the state of the law to cut back production. Indeed, individual US
investment in databases must have been climbing over the period, because the number of
players in the industry increased only 10% over the period.
Does strong database protection lead to increased incentives,
and increased database production?
According to the Directory, online database production may not have increased as
a result of the Directive.16 Indeed, the Directive may have had no effect at all- production
of online databases stayed nearly flat after the directive was implemented. In some
countries, the number of new databases created each year fell after implementing the
Directive. And the correlation between implementation and new database production is
statistically insignificant – the presence or absence of implementing legislation explains
just over 1% of database production.
Some researchers, writing in 2001 and thus without the benefit of later data, have
written that the Directive has, in essence, failed.17 They argue that, aside from a brief
spike in total database production immediately after implementation, preceeded by a
commensurate decline in production while firms anticipated the new legislation, the
Directive has had no obvious effect and database production has fallen to pre-Directive
Band, supra note 6, citing Martha E. Williams, “The State of Databases Today: 1998”, in Erin E.
Holmerberg, ed., Gale Directory of Databases, p. xvii (September 1997).
16
The Directory was queried electronically, using Westlaw’s interface, for databases identified as
originating in each country in each year between 1994 and 2003, inclusive. Only the databases that were
new to the directory in each year were reported. The total number of responses to each query were then
plotted. Thus, each data point represents the number of new databases in each country in each year. The
queries counted only database creation; no attempt was made to count the number of databases that ceased
being available in any year.
17
Stephen M. Maurer, P. Bernt Hugenholtz, Harlan J. Onsrud, “Europe’s Database Experiment,” Science
Vol. 294, Num. 26 (October 2001). For more concrete data, see Stephen M. Maurer, Suzanne Scotchmer,
“Across Two Worlds: Database Protection in the US and Europe,” in J. Putnam, Intellectual Property and
Innovation in the Information-Based Economy, forthcoming.
15
6
levels. This, they take it, is empirical evidence of the reasons not to adopt strong
protection.
On the other hand, as has been pointed out elsewhere this is at best negative
evidence.18 The lack of correlation tells us only that other factors affected database
production more than new protection measures, not that the measures had no positive
effect whatever. Moreover, the United States observed a similar downward trend in
online database production over the period; perhaps something about database usage or
technology changed during the period to make online publication of databases
unattractive. (See Figure 1).
In fact, the same period showed exponential growth in the size and consumption
of many existing databases, both in the United States and in Europe.19 At least in
biological databases, new records were being inserted and the rate of incoming queries
was also climbing. Thus data was still being produced, and the European trends appeared
to follow American trends in data usage and production.
Trends are not quantities, however. For the entire period measured, U.S. online
database production outpaced all of Europe by a factor of nearly 2.5:1, even though it fell
in America at the same times and rates as in Europe. American dominance of database
production cannot be explained by the incentives given to creators because American
protection of database rights is much weaker than the Directive.
See Hughes, supra note 1. Hughes notably calls the current state of evidence either “a ringing
endorsement of the status quo,” or a recommendation that changes should be reactive to observed
phenomena.
19
See Mark Gerstein, Dov Greenbaum, “A Universal Legal Framework as a Prerequisite for Database
Interoperability,” Nature Biotechnology Vol. 21, Num. 9, pp. 979-82 (September 2003).
18
7
Online Database Production
160
140
120
100
Germany
United Kingdom
United States
Europe
80
60
40
20
0
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
Figure 1
Is robust database protection the only or best way to drive new
database production?
If not protection, then what explains America’s database productivity? The
importance of this question to policy-making ought not be gainsaid: if something other
than high levels of protection may incent database creation, and may be more efficient in
doing so, it ought to be explored and its harmful side-effects compared to those of the
current database protection proposals.
Certainly other methods of driving database production could be, and have been,
successfully employed. For instance, some databases have been considered public sector,
and thus left completely unprotected in the United States. The government has explicitly
financed these databases, and the resulting spin-off revenue has paid for them many times
8
over.20 This model of database production has proven efficient for at least some kinds of
databases (in particular, databases that require government monopolies to generate, or
that are best generated in this way, like meteorological and topographical databases).
Moreover, several other models for providing incentives to create have been proposed,
and their theoretical qualities tested.21
If increased database protection has not benefited the broader public by increasing
database production, it has had some effect. Strong database protection means that
database creators must expend a great deal of capital; it is no longer possible to legally
repurpose the data contained in existing databases in order to cheaply create a new
database. Database protection is capital-rich company protection. As such, it runs the
risk of actually decreasing database concentration. Perhaps this helps explain some of the
overall decline in database production in Europe, although the drop-off in American
growth at the same time is striking, and suggestive of broader sources.
The NautaDutilh study.
The Directive demands that it be periodically reviewed. The reviewing process
itself creates more data, and more empirical evidence. The first mandatory deadline
described in the Directive has passed, and the report was not forthcoming. However, a
See Peter Weiss, “Borders in Cyberspace: Conflicting Public Sector Information Policies and their
Economic Impacts,” Summary Report, National Oceanic and Atmospheric Administration, National
Weather Service, (February 2002). Weiss compares American public sector databases, which are
distributed for little or even no charge, to some European databases that are distributed under restrictive
terms at higher cost and concludes that the American system leads to more and better database production,
and higher returns on the databases so developed.
21
In the patent domain, see Nancy Gallini, Susan Scotchmer, “Intellectual Property: When Is It the Best
Incentive System? (preprint),” Innovation Policy and the Economy Vol 2, Adam Jaffe, Joshua Lerner and
Scott Stern, eds, MIT Press, pp. 51-78, and also forthcoming in Legal Orderings and Economic Institutions,
F. Cafaggi, A. Nicita and U. Pagano, eds., Routledge Studies in Political Economy. Available at
http://socrates.berkeley.edu/~scotch/G_and_S.pdf. In copyright Maurer and Scotchmer note that
technological and market mechanisms have protected American databases. Stephen M. Maurer, Suzanne
Scotchmer, “Database Protection: Is It Broken and Should We Fix It?,” Science Vol. 284, Num. 5417, pp.
1129-30 (May 14, 1999).
20
9
review is forthcoming, based primarily on the results of a survey conducted by a private
firm under contract. No discussion of the existing empirical evidence could be complete
without an exploration of this survey.
The first thing to note about the survey is that its results are statistically
insignificant, because it had only 45 respondents, and they were not randomly chosen.
This does not remove from it all value, but it ought to color one’s reading of the final
interpretations of results. The second thing to note is that the respondents self-selected
for a certain viewpoint. Of the 45, 10 identified themselves as database publishers, and
none identified themselves as consumer protection representatives. The mode
represented companies turning over more than 200 million EUR each year, and the
majority made at least 50 million EUR, so the answers are all bent towards the interests
of large, high-income rights holders (none had less than 50 employees, and the mode had
more than 2000). The most important deficiency of the study, however, is that it failed to
ask quantifiable questions about the actual effect of the Directive on decision making, but
instead required the respondents to assess broad claims about the effect on the decisions
of everyone within the system. For instance, rather than asking whether the database
owner had sold their database outside of their home country, the survey instead asked that
each respondent mark either “I agree” or “I disagree” to statements like, “By eliminating
the differences existing between the Member States’ legislation…, the Directive has
[had] a positive effect on the free movement of database-related goods or services within
the Community.”
What is astonishing about the survey is that, despite these obvious biases, nearly
twice as many respondents felt that the Directive created an unbalance between the rights
10
of rightsholders and users. Nearly half of those who responded to the question thought
the definition of a database in the Directive was either too broad or too uncertain. None
thought it too narrow. A healthy majority thought the substantial investment threshold
for so-called sui generis protection was too uncertain. Respondents were split down the
middle on whether the creation of sui generis protection was satisfying. Most thought
such rights too broad. Many noted that the permitted exceptions were too narrow.
Conclusion
The debate over whether to offer databases additional protection beyond
copyright has focused on the theoretical economic arguments on both sides of that issue.
The arguments have been suspiciously devoid of evidence. In fact, evidence does exist,
but it is both hard to find, and difficult to meaningfully analyze.
The current state of American database production suggests that there is no
problem to be corrected. The number of companies producing databases is growing
moderately. The number of databases those companies are producing is growing much
more quickly, and the number of entries in those databases is exploding exponentially.
Moreover, the industry has experienced handsome revenue growth, has created new jobs,
and is poised to grow even more in the future. All of this despite the fact that America
does not offer strong, legislated database protection. Current database publishers rely on
a combination of self-help mechanisms to protect their investments, and the mechanisms
work.
The situation in Europe is not as happy, despite the strong database protection that
exists as a result of the European Database Directive. Database growth has been
moderate, at best, and not much better if better at all than it was before the Directive.
11
Moreover, despite the assertions of proponents of strong protection, even database
producers are unhappy with many unforeseen aspects of the Directive and its effects.
That said, it would be premature to conclude from the evidence that database
protection simply does not work to spur new production. Other explanations cannot be
ruled out, including the simplest that complex database protection schemes will take a
long time to have any appreciable effect. Thus, if there is a problem then database
protection might be a way to solve it.
There are other good ways to solve it, however, and they have not been explored
significantly. Public sponsorship of database production, including direct contract
payments for certain kinds of work, have been remarkably successful in producing a
robust and high-quality database industry. Even better, publicly funded and generated
databases may be made available for reutilization at little or no cost, which might then
drive other productive activities. Because data is an input to most modern production,
systems that spur data production and dissemination have much to recommend them.
12
Appendix: Data and Analysis
The Gale Directory was queried on Westlaw. Each data point was created by searching
for: ‘CN(“country”) AND YR(year)’. The total number of responses to the query for
each country and year were compiled into the following table. The highlighted entries
indicate the years in which the legislation incorporating the European Database Directive
took effect.
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Luxembourg
Netherlands
Portugal
Spain
Sweden
United
Kingdom
United
States
1994
0
4
0
0
2
20
0
0
0
0
4
0
0
0
9
1995
1
2
4
0
3
13
0
0
0
0
1
0
0
1
16
1996
1
3
1
3
2
5
0
0
0
0
1
0
0
2
8
1997
2
0
1
3
3
10
0
0
1
1
5
0
0
1
28
1998
0
1
0
1
1
9
0
1
0
0
2
0
0
0
48
1999
2
12
1
0
1
7
0
1
1
0
0
0
1
0
74
2000
0
0
2
1
1
7
0
0
2
1
1
0
0
0
14
2001
0
0
0
0
0
1
0
0
3
0
0
0
0
1
2
2002
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
2003
0
0
0
0
0
2
0
0
0
0
0
0
0
0
1
86
125
111
145
122
136
85
32
7
5
The data was then analyzed for correlations between implementation of the European
Database Directive and the number of new databases listed in the Gale Directory. The
correlation was calculated to be 0.012644, or about 1.3%.
The United Kingdom presents the obvious outlier among European countries, and
Germany was highly productive over the period as well. On average, Europe produced
35.87 new databases in the Directory before implementing the Database Directive, and
38.08 after, a 6% boost. However, this is largely due to the UK; without United
Kingdom production, Europe produced 21.62 databases per year on average before
implementation, and only 11.47 after, a decline of 47%. The United States added an
average of 85.4 new online databases each year over the period.
13