Photo by GraphicMama-team via Pixabay

In 2015, the Program on Information Justice and Intellectual Property (PIJIP) set out to conduct empirical research on the impact of copyright exceptions. We felt that there was a shortage of papers demonstrating the benefits of exceptions for users of protected works. We soon realized that information about the changes to copyright law over time – which would be especially useful for empirical studies – was lacking. There have been many studies describing differences in copyright exceptions between countries, such as those by Seng, Crews, Hilty & Nérisson, and the WIPO Secretariat.  However, these studies are static, capturing differences in each country’s laws at one moment in time. We wanted a dataset allowing us to compare laws across countries and over time, and we set out to develop one.

PIJIP reached out to professors and attorneys in the Global Expert Network on Copyright User Rights. We held a series of workshops during which we gradually developed a survey that would cover a wide range of copyright exceptions. Legal scholars from our network completed the survey, which we used to construct the User Rights Database. Sean Flynn and I first introduced the database in white papers in 2017 and 2018, which focused on the openness of the various copyright exceptions. Last year, I wrote a new paper that more formally introduces the data, and shows how subsets of the data can be used to measure changes to particular types of copyright exceptions.

This post provides an overview of the paper, A Novel Dataset Measuring Change in Copyright Exceptions, which is available here. The survey instrument and the data are here.

A first look at the data

PIJIP’s User Rights Database currently contains data on copyright exceptions in 24 countries over 27 years.[1]Thirteen countries in the set are High Income Countries as classified by the World Bank, and the rest are Middle Income. The distinction is relevant because legal academics have argued that copyright laws in less wealthy countries tend to have weaker exceptions than copyright laws in wealthy countries – despite the TRIPS Agreement’s flexibilities allowing countries policy space to permit certain unauthorized uses Table 1 lists the countries by income group.

Table 1. Countries in the User Rights Database

High Income Countries Middle Income Countries
Australia

Canada

Chile

Finland

Japan

Korea

Netherlands

Portugal

Singapore

Slovakia

Switzerland

Taiwan

United States

Argentina

Botswana

Brazil

China

Colombia

India

Mexico

Peru

South Africa

Ukraine

Vietnam

 

Our survey contains 129 questions, grouped into 20 categories based on different types of copyright exceptions (i.e. the right to quote works without authorization). Table 2 shows a list of all of the categories of questions.

In each category, the survey asks respondents whether their country’s copyright law included a relevant exception each year from 1970 to present. It then asks a series of follow-on questions relevant to that particular exception. These vary from one category to the next, but they generally include whether it applied to unauthorized uses for any type of work, any purpose, any type of user, and whether it could be used for commercial purposes. Some categories also include additional questions relevant only to the category at hand. For instance, the category for libraries includes the question of whether the exception allowed unauthorized reproduction to provide copies for other libraries. For each of these follow-on questions about each user right, the survey asks for the year in which any change occurred. The complete survey is included in the working paper as Appendix 1.

In order to gauge the level of uncertainty surrounding specific copyright exceptions, respondents answered these questions on a four-point scale running from “Clearly Not Included” to “Clearly Included.” Ambiguity could exist due to differing interpretations of legal texts, or due to judicial interpretations of laws that predate legislative change.  The answers were coded 0 to 3, allowing one to observe the strength of each exception in each country’s law.

Table 2. Survey Categories and Descriptive Statistics

Category Mean St. Dev.
General Exception, Including Fair use 1.01 1.32
Quotation 2.09 1.18
Education 1.64 1.30
Research 1.53 1.33
Personal or Private Uses 1.80 1.27
Computer Programs 1.43 1.33
Databases or Other Compilations of Non-Original Facts 0.93 1.24
Text- and Datamining 0.46 0.92
Library Rights 1.81 1.33
Disability Access 1.28 1.33
Transformative Use 0.69 1.06
Parody and/or Satire 1.31 1.25
Incidental Inclusion 1.17 1.34
Panorama Right 1.77 1.33
Orphan Works 0.75 1.18
National Government Works 1.38 1.34
Exhaustion of Rights 1.32 1.31
Safeguards from Secondary/Intermediary Liability 0.62 1.14
Temporary Copies for Technological Processes 1.00 1.31
Protection Against Supremacy of Contracts 0.64 0.93

 

Table 2 shows the average score by category of question. One can see that the countries in this set had stronger copyright exceptions in some categories than others. For example, the mean scores for exceptions protecting quotation, education uses, and personal or private uses are each above 1.5. Their standard deviations are relatively low (compared to most of the other categories), indicating there is less variation from one county to the next. This makes sense when one considers that copyright exceptions for quotation, education, and personal or private use are well established under international copyright law. The Berne Convention explicitly recognizes copyright exceptions for quotation and education. Most countries have allowed some sort of personal use exception for a long time.

On the other hand, copyright exceptions that apply to ICT technologies tend to be weaker and more varied. The surveyed countries are less likely to have protections for text- and data-mining, databases, transformative uses and safeguards for intermediary liability in their law. When national laws do include these types of laws, the exceptions tend to be more restricted in terms of the type of uses they permit. The average scores for each of these types of copyright exceptions are below 1.0.

 

Measuring the strength of subsets of copyright exceptions

Subsets of the data can be used to compare different types of copyright exceptions between countries and over time. To demonstrate, my paper presents two overlapping thematic subgroups of exceptions based on the type of user activity protected, and presents indices for each. The first index measures the strength of exceptions related to ICT technologies, and the second measures the strength of exceptions related to education.

Table 3 shows the categories of copyright exceptions relevant to each subgroup. (Appendix 2 of the working paper provides the full list of questions.)

Table 3 Copyright Exceptions for Two Types of Uses

Exceptions related to Internet Communications Technologies (Tech) Exceptions related to Education (Edu)
General Exception

Quotation

Research

Personal Or Private Uses

Computer Programs

Databases Or Other Compilations Of Non-Original Facts

Text And Data-Mining

Transformative Use

Safeguards From Secondary Liability

Temporary Copies For Technological Processes

Supremacy Of Contracts

General Exception

Education

Research

Personal or Private Uses

Library Rights

Exhaustion of Rights

 

Figure 1 is a graphic presentation of the mean values of Tech and Edu each year by income group. For each year,

  • tech = mean of the category scores for 52 questions within the eleven categories related to Internet Communications Technologies (ICT) in Table 3
  • edu = mean of the category scores for 31 questions within the six categories related to educational uses in Table 3

It demonstrates three things. First, the mean value of both indices has gradually increased over time, indicating that copyright exceptions have slowly grown stronger. This is true for both the high-income and middle-income countries. Second, the high-income countries in the set have consistently had stronger copyright exceptions than the middle-income countries, supporting the assertions that developing countries have not taken full advantage of TRIPS flexibilities for copyright protection. Third, the gap between copyright exceptions in more and less wealthy countries grew towards the end of the period. This is especially the case for exceptions that benefit technology firms.

Figure1: Average Index Scores for High- and Middle-Income Countries, by Year

 

 

Testing covariates

My paper uses panel regressions (with fixed effects for countries and years) to show the relationships between the strength of copyright exceptions and other economic variables. The main results are as follows. (further details in the paper):

  • Exceptions related to ICT technologies tend to be stronger (on average) in countries with larger ICT sectors, and exceptions for educational uses are stronger in countries with higher educational attainments.

 

  • Exceptions related to both ICT technologies and educational uses tend to be weaker (on average) in countries where the copyright industries have a higher relative share of GDP.

 

  • Within each country, there is a positive correlation between both indices and GDP per capita, though it is only significant for Tech. Rather than describing the difference between high- and middle-income countries, this finding implies that as economic output increased in each country, counties tended (on average) to strengthen copyright exceptions for ICT technologies.

 

Conclusion

This post has given an overview of my working paper, which presents PIJIP’s User Rights Database, a novel dataset designed to measure changes in 24 nations’ laws on copyright exceptions over time. It demonstrated variation in exceptions by category, by countries’ income group, and over time. The User Rights Database is unique among sources of information on copyright exceptions because others are static. It is my hope that researchers can use the dataset to add to the relatively small body of empirical research in this area.

Finally, the User Rights Database is an ongoing project. As we find new survey respondents with the expertise needed to answer detailed questions about the history of their countries’ copyright law, and the willingness to spend time on the survey, the database will gradually expand to cover additional countries.


[1] Some countries in the dataset have data going back to 1970. However, there is annual data for most countries in the dataset from 1990 to 2016.


_____________________________

Kluwer IP Law
This page as PDF

Leave a Reply

Your email address will not be published. Required fields are marked *