An ‘Algorithmic Links with Probabilities’ Concordance for Trademarks with an Application Towards Bilateral IP Flows

• Author: Travis J. Lybbert,Nikolas Zolas,Prantik Bhattacharyya
• Published date: 01 June 2017
• Date: 01 June 2017
• DOI: http://doi.org/10.1111/twec.12382

An ‘Algorithmic Links with Probabilities’

Concordance for Trademarks with an

Application Towards Bilateral IP Flows

Nikolas Zolas

1

, Travis J. Lybbert

2

and Prantik Bhattacharyya

3

1

Center for Economic Studies, United States Census Bureau, Washington, DC, USA,

2

Agricultural &

Resource Economics, University of California –Davis, Davis, CA, USA, and

3

Computer Science,

University of California –Davis, Davis, CA, USA

1. INTRODUCTION

IN the contemporary global economy, trademarks (TMs) play an important role in a wide

array of industries and sectors and shape the competitive landscape of many diverse mar-

kets. Although reliance on TMs certainly evolves with structural changes and economic devel -

opment, the economic importance of TMs is as apparent in developed countries as it is in

emerging and even developing countries. Despite these realities, economists and policy ana-

lysts alike have been unable to conduct careful empirical analysis of TMs in the modern econ-

omy because TM data and economic activity data are organised differently and can therefore

not be analysed jointly. In this project, we aim to remedy this incompatibility by building a

bridge between TM and economic data that enables these data to ‘speak to each other’.

This work is motivated by the paucity of rigorous empirical research into the relationship

between TMs and economic activity, which is due to two primary facts. First, the competitive

and strategic considerations that shape whether and how firms rely on TMs to build brands

and differentiate their products and services differ dramatically across industrial sectors. This

implies that any empirical analysis should either focus on TM activity in a particular sector or

otherwise allow for substantially different empirical relationships between TMs and economic

activity across sectors. Second, while TM data are available from more and more countries

and economic data are widely available at a high resolution of industrial sector or product cat-

egory, merging these data by linking the goods & services (GS) covered by a TM to sectors

or products is difficult and –to date –has been very limited. This presents a serious con-

straint on matching TM and economic data at a useful level of resolution and in a robust and

reliable way. In conjunction with the first fact, this severely limits the kinds of empirical

research that are possible in this area.

We develop an algorithmic approach we call ‘algorithmic links with probabilities’ (ALP)

matching to explicitly link TM and economic data via standard, widely used product and indus-

try classification systems such as the Standard International Trade Classification (SITC), Inter-

national Standardized Industrial Classification (ISIC), Harmonized System (HS), and North

Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent

the views of the US Census Bureau. All results have been reviewed to ensure that no confidential infor-

mation is disclosed. We thank members of the US Patent and Trademark Office (USPTO), World Intel-

lectual Property Organization (WIPO) and participants of the USPTO-NYU Trademark Workshop for

comments and assistance with the data. Portions of this paper were originally prepared as a report for

WIPO with support from WIPO.

©2016 John Wiley & Sons Ltd

1184

The World Economy (2017)

doi: 10.1111/twec.12382

The World Economy

American Industry Classification System (NAICS). As a key benefit to this approach, these

class-level ALP concordances implicitly reflect differences in TM usage across economic sec-

tors –and therefore link TMs to economic activity according to predominant TM-use patterns.

This ALP matching approach, which has been used similarly to concord patents to

economic data (Lybbert and Zolas, 2014), enables researchers to map TM data directly into

trade or industry categories in order to create measures of TM-use intensity that are compara-

ble across countries and over time and to empirically model the determinants of international

TM flows and the economic effects of TMs. Together with similar ALP concordances

designed for mapping patents into the same economic classification systems, these new data

tools open up broader possibilities to jointly analyse TMs and patents. Given how much intel-

lectual property strategies vary from industry to industry and given the interdependence that

is often evident in the use of these two important forms of intellectual property, the ability to

combine patent, TM and economic data by industry into a single analysis is particularly

potent. Such joint analysis would reflect the inherent heterogeneity in TM usage across sectors

described above and would ultimately improve our understanding of the relationships between

intellectual property and the value of production of both goods and services domestically and

the value of goods traded internationally. Analyses such as these could not only improve our

ability to model and understand how TMs fit into the contemporary global economy generally,

but would also serve as a platform for addressing a host of policy-relevant research questions.

2. BACKGROUND

Trademark filings have expanded rapidly in recent decades. As described by the 2012

World Intellectual Property Indicators report (WIPO 2012), total TM applications worldwide

more than doubled between 1995 and 2011, with more than 4.2 million applications filed in

2011. Much of this growth was driven by TMs filed in and by emerging economies, with

China accounting for nearly half of the overall growth between 2004 and 2011 (46.9

per cent). What is somewhat surprising about this growth is that while overall TM output has

increased dramatically, the level of foreign TMs (i.e. those registered outside the registrant’s

home jurisdiction), has more or less stayed flat over this same time period, despite the dra-

matic increase in trade and other forms of transferred intellectual property, such as patents.

Institutional innovations have facilitated these internationally filed TMs. Specifically, the

Madrid Protocol became operational in 1996, making it much easier for TM owners to apply

for international registrations in countries that have joined this protocol.

1

Using data from the WIPO, we provide additional perspectives on these trends. We focus

mainly on foreign TM registrations (so-called exported TMs) and consider how these exported

TMs flowed from and to different country incomes classes during the past two decades. We

classify countries into income classes using the World Bank high, middle and low-income cat-

egories. Table 1 shows average annual TM registrations sent to and from these different

income categories over the years 1994–2011. While high-income countries filed on average

10 times more TM registrations than middle-income countries and more than 100 times more

1

This protocol materialised from the original Madrid Agreement, which first entered into force in 1892

as a means for international TM registrations and had 56 member countries at the time the protocol was

agreed upon. Today, there are 90 member countries in the Madrid Protocol, allowing TM holders to

extend the jurisdiction of their TM to anyone of these countries at any time during the life of the TM.

©2016 John Wiley & Sons Ltd

ALP CONCORDANCE FOR TRADEMARKS 1185