Innovation with and without patents Josef Taalbi1 1Department of Economic History Lund University

2025-05-05 0 0 1.73MB 50 页 10玖币

侵权投诉

Innovation with and without patents

Josef Taalbi,1∗

1Department of Economic History, Lund University

∗E-mail: josef.taalbi@ekh.lu.se.

A long-standing discussion is to what extent patents can be used to monitor

trends in innovation activity. This study quantiﬁes the amount and quality

of information about actual innovation contained in the patent system, based

on 4,460 Swedish innovations (1970-2015) that have been matched to interna-

tional patents. The results show that most innovations were not patented and

that among those that were, 43.9% of all innovations, only a fraction can be

identiﬁed with patent quality data. The best-performing models identify 17%

of all information about innovations, equivalent to an information loss of at

least 83%. Econometric tests also show that the fraction of innovations re-

sponding to strengthened patent laws during the period were on average 8%

percent. The overlap between the patent and innovation systems is hence more

modest than often assumed. This accentuates the need to, alongside patents,

develop versatile approaches in order to induce and monitor various aspects

of innovation.

arXiv:2210.04102v1 [econ.GN] 8 Oct 2022

Introduction

Innovation is widely viewed as essential for achieving long-run economic development and

sustainability. Measuring and monitoring innovation is therefore key for our understanding of

trends in technology and structural transformation over time and between countries. Innovation

is typically measured through various indicators covering different stages of the innovation

process [1], including scientiﬁc output, R & D expenditures and patents, or compound indices

of these [2, 3, 4].

Patents remain the most widely used innovation indicator and for good reason. Patent

records are publicly and readily available, contain detailed information about inventors and

inventing ﬁrms globally and for long time periods. In addition, patent records offer ways of

assessing the value of patents through patent citation counts [5, 6, 7], and the insights into

knowledge ﬂows that can be gained from analysis of patent citations are unparalleled [6].

There are important sources of discrepancies, however, between innovations and patents,

and the extent to which the patent and innovation systems overlap is not settled [8, 9]. The

question of how much the patent and innovation systems actually overlap also haunts policy

discussions. The patent system has come under ﬁre for being inefﬁcient [10, 11, 12], distort-

ing incentives or blocking innovation [13, 14], while the number of innovations affected by

patents or patent laws has been argued to be low or uncertain [15, 16, 8, 17]. Gaining a better

understanding of this overlap is therefore key also for policy debates.

This study presents evidence about the amount and quality of information on innovation that

is possible to identify within the patent system. In doing so, this study addresses methodological

issues and supplies evidence for broader discussions about the relationship between the patent

and innovation systems. This effort is based on a literature-based innovation output (LBIO)

database [18, 19] containing 4,460 commercialized innovations in Sweden, one of the world’s

highest ranked innovative economies [4]. These innovations, commercialized between 1970

and 2015, were linked to 13,561 patents across various national and supra-national patent of-

ﬁces, through manual and machine-learning-assisted searches in Google Patents. The matching

methodology is detailed in the supplementary materials.

This data enables examination of the information overlap between patents and commercial-

ized innovations from multiple angles. This study views the problem of measuring innovation

as analogous to information transmission from a source to a receiver through a noisy channel

(Fig. 1). The key question is how much information the receiver (patent analyst) has about the

source (innovations). This study phrases this question as follows: what fraction of innovations

can be correctly identiﬁed by a patent analyst, based on data available within the patent system?

This fraction is circumscribed by three factors (Fig. 1). First, not all innovations are

patented, but some fraction ρthat is determined by property laws and appropriability strate-

gies. Previous research proposes varying estimates of the percentage of innovations that are

patented, from 9.6% [8], 36% [20] to almost half [21]. A second important issue is that patents,

strictly speaking, reﬂect invention, which may, or may not, lead to innovation, viz. new com-

binations that are commercialized or otherwise come into economic use [22]. Patenting is also

often the outcome of strategic decisions to protect intellectual advances, rather than reﬂecting

innovation activity [23]. Therefore, to weed out less important or less valuable patents, the usual

strategy is to use patent citations. However, while some studies suggest patent citations to be

a good measure of economic value [5, 6, 7], others ﬁnd that patent citation counts are “noisy”,

heterogeneous over time, across sectors and countries [24, 25, 26, 9, 27]. Such quality-adjusted

indicators must balance two aspects: the amount of actual innovations (true positives) that are

captured, the recall α, and the fraction of true positives among all patents identiﬁed, the preci-

sion β. The information about innovations in the ﬁnal patent selection is then deﬁned by the

fraction of innovations covered and the precision of the selection: ρ×α×β. This measure is

further motivated in the supplementary materials.

ρpatented recall α

with precision β

Innovations Patent

data

Identiﬁcation

patented

innova-

tions

Skype

Spotify

NMT

SBR

Skype

Spotify

Patent 1

Patent 2

Patent 3

Skype

Patent 1

Patent 2

ρ= 1/2α= 1/2,

β= 1/3

Figure 1: (a) Innovations and patents viewed as information transmission through a noisy chan-

nel. A fraction ρof Ninnovations enter into the patent system. The patent system also contains

noise in the form of non-commercialized inventions that can be reduced through patent quality

measures. The quality of the information about innovations depends on the fraction of true

positives identiﬁed (recall α), and the probability that a patent identiﬁed is truly an innovation

(precision β). (b) Example. If asked to name the patented innovation, the patent analyst would

on average be able to correctly identify 1 correctly a third of the times. The information about

the original source, as a fraction of the total, is ρ×α×βor 1/12 in the example.

Results

Patent propensity We turn ﬁrst to the propensity to patent (ρin Fig. 1). The main results

are given in Fig. 2 and further detailed in Table S1. The results show that 43.9 percent of

all innovations, launched in 1970-2015, were patented in at least one patent ofﬁce, whereas

(a)

(b)

Figure 2: (a) Patent propensity by patent ofﬁce. (b) Total patent propensity, by commercial-

ization year 1970-2015, (c) EPO, USPTO, Sweden, Japan, (d) Patent propensity across sectors

(ISIC Rev. 3), all patent ofﬁces. Note: Results not given when sectoral counts are below ﬁve.

patenting propensity to any one single patent ofﬁce was highest for the Swedish and US patent

ofﬁces. Combining data from two or three patent ofﬁces however sufﬁces to capture a near-

complete set of all patented innovations (Fig. 2a).

Piercing below the aggregate reveals stark differences over time and across sectors. The

patent propensity to the Swedish Patent Ofﬁce has increased from a low level of 13.9% in 1970

to 57.7% in 2000, followed by a decrease back to 21.1% in 2015. The patent propensity to

USPTO increased from 26.6% in 1970 to 49.0% in 2000, followed by a decrease back to 25.3%

in 2015 (Fig. 2c). These results are in line with the emergence of a pro-patent era in the late

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

InnovationwithandwithoutpatentsJosefTaalbi,11DepartmentofEconomicHistory,LundUniversityE-mail:josef.taalbi@ekh.lu.se.Along-standingdiscussionistowhatextentpatentscanbeusedtomonitortrendsininnovationactivity.Thisstudyquantiestheamountandqualityofinformationaboutactualinnovationcontainedinthepatent...

展开>> 收起<<

Innovation with and without patents Josef Taalbi1 1Department of Economic History Lund University.pdf

共50页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Innovation with and without patents Josef Taalbi1 1Department of Economic History Lund University

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: