Chapter 13 Innovation Statistics

Giuliana Battisti and Paul Stoneman

What is innovation and why do we measure it?

Why do we not just have one measure of innovation?

What are the existing sources available, and how accurate are they?

What is missing from innovation statistics?

How far back in time can we measure innovation?

13.1 Introduction

The definition we use is from the OECD’s glossary of statistical terms.

innovation

The implementation of a new or significantly improved product (good or service), or process, a new marketing method, or a new organisational method in business practices, workplace organisation or external relations. This new and/or improved product or process differs significantly from the unit’s previous products or processes, and has been made available to potential users (product) or brought into use by the unit (process).

On 1 October 1963, Harold Wilson, at that time the leader of the UK’s political opposition, made a famous speech at the Labour Party conference in Scarborough. For the first time a leading national politician chose to focus on the role of innovation as the engine of growth for the economy. The British people were, he said, “accepting as part of their everyday life things which would have been dismissed as science fiction a few years ago”.

He went on: “This scientific revolution” he said, “is making it physically possible … to achieve for the whole people better living standards than those enjoyed by tiny privileged classes in previous epochs”.

A year later, Wilson’s vision of how new ideas could become products and services that improved everyday life – what he called the “white heat of technology” – had helped him to become prime minister, a job he held until 1970.

Today we more easily accept that innovation drives long-run economic growth, nationally and globally. It is also a prime determinant of the competitiveness of nations (and thus export performance), the competitiveness of firms (and thus firm output growth, employment and profits), and the enhancement of our wellbeing.

As such, innovation is at least as important as any other production activity, and has to be fully understood, monitored, managed and measured. However, we now take a much wider view of what innovation encompasses than we did when Wilson made his speech. Innovation can also mean changes in management methods and techniques, product design and, in many sectors, changes in product aesthetics too.

Innovation plays a role in economic growth and development that is too important for it to be left to occur by chance. Policymakers need to identify where to intervene usefully in the innovation process, and they can do that only if they gather statistics on the ways firms innovate, and how successful they have been.

William Baumol, an economist who specialised in innovation, put it like this:

…firms cannot afford to leave innovation to chance … [they] are forced by market pressures to support innovation activity systematically … The result is a ferocious arms race among firms in the most rapidly evolving sectors of the economy, with innovation as the prime weapon.¹

To create policies to encourage innovation, we first have to measure it, and its effect. This is problematic. It is difficult to define innovation, but even harder to measure it – not least because what we measure depends on what we define.

13.2 What is innovation?

The simplest definition of innovation that we could have used is that it is “the successful exploitation of new ideas”, the definition used by the Innovation Unit of the Department for Business, Innovation and Skills in the UK. We use this narrow definition to begin our investigation in section 13.3.

The use of the term in modern economics goes back to the work of Joseph Schumpeter, the first economist to analyse in detail the impact of innovation on technological change and growth. Later we want to learn about the effectiveness of innovation policy, so we need to understand, as Schumpeter did, that innovation has three stages. Before implementation there must be an idea, and after implementation there is the value that the idea generates.

Schumpeter’s ideas are set out in:

Schumpeter JA (1934), The Theory of Economic Development: an Inquiry into Profits, Capital, Credit, Interest and the Business Cycle, New Brunswick, NJ: Transaction

He considered that innovation encompassed new products, new processes, new materials, new management methods and new markets, and involved three interrelated stages:

1. Invention: the generation of new ideas
2. Innovation: the translation of new ideas into new products, and production and business processes
3. Diffusion: the spread of the latter across their potential uses.

We will call Schumpeter’s three-step analysis the wider concept of innovation, and we deal with that in sections 13.4, 13.5 and 13.6.

Note that both definitions use the phrase “new ideas”. There are three main concepts of newness:

• New to the world: something that has not previously existed
• New to the country: something that has not previously featured in the economy under consideration
• New to the individual economic unit: something that has not previously been part of the activities of the industry, firm, household or other economic unit being considered.

Instinctively we think of innovation as something that is new to the world. That would be a very incomplete picture of the innovative activity we want to quantify. Most profit-generating innovations in firms or households will not even be new to the industry in which they are applied, let alone new to the world.

We will start with the challenges of the measurement of the narrow concept of innovation and look at the available statistics and what they show.

13.3 Measuring narrow innovation

13.3.1 What we include, and why

The definition and measurement of the narrow concept of innovation has been largely guided to date by what is known as the Oslo Manual. The latest version, published in 2018, defines business innovation as

…a new or improved product or business process (or combination thereof) that differs significantly from the firm’s previous products or business processes and that has been introduced on the market or brought into use by the firm.

More formally entitled Guidelines for Collecting and Interpreting Innovation Data, the “Oslo Manual” was first published by the Organisation for Economic Cooperation and Development (OECD) in 1992. The 1992 edition was mainly concerned with measuring (technological) innovation within the business enterprise sector and was restricted to manufacturing.

The second and third editions broadened the scope to market services and the concept of innovation to include marketing and organisational innovations. We are now on the fourth (2018) edition.

Innovation statistics compiled to date mostly use the guidelines in the 2005 manual. This may change when the 2018 edition is employed.

This distinguishes between innovation as a process and innovation as an outcome. One may think of the innovation process as an (economic) activity that uses resources in order to generate innovation outcomes. This process itself may be studied in order to explore its productivity (output per unit of input), for example. Innovation as an outcome is essentially an attempt to measure the outcomes from the process, and it is these outcomes that the manual refers to as innovation. Both aspects are relevant to both the wide and narrow concepts of innovation. The distinction suggests that it may be advantageous to separate measures of innovation relating to inputs to the process from those relating to outputs from the process.

We can measure innovation at the macro (national or international) level or at the micro (industries, firms, households) level. But, when we do this, aspects of the 2018 Oslo Manual definition raise problems – most importantly, the number of changes that can be recorded as innovation.

In the guidelines, the minimum requirement for an innovation is that the product or business process must have one or more characteristics that are “significantly different” from those contained in the products or business processes previously offered by or used by the firm. These characteristics must be relevant to the firm or to external users. Product innovations must provide “significant improvements” to one or more characteristics or performance specifications.

There is a very long list of relevant functional characteristics which includes:

• quality
• technical specifications
• reliability
• durability
• economic efficiency during use
• affordability
• convenience
• usability
• user friendliness
• appearance
• affordability
• financial convenience.

That’s not all. An improvement to or addition of a new function can also be combined with a loss of other functions, or a decline in other performance specifications. So poorer quality, cheaper products can now be counted as innovations.

Note also that “significant” is not explicitly defined. An obvious gauge of significance would be market impact – an innovation that makes a large market impact would be considered as more significant than one that fails.

But the Oslo definition focuses on the process, and that would be an outcome. So a change does not need to be a commercial, financial, or strategic success, either at the time of measurement, or ever! A product change can fail commercially, or a business process change may require more time to meet its objectives. Around 40% of product innovations fail to deliver commercial success. By this definition, innovation does not necessarily improve the market position or financial performance of the firm, even if users benefit from it.

Nor does the definition of an innovation require it to have a positive value for society or consumers (outcomes, again), as distinct from the firm. An innovation can lead to a significant boost in the financial performance of the firm while providing fewer benefits to consumers than other offerings from the same firm, or its competitors. This means an innovation, by this definition, can create safety, health or environmental problems.

13.3.2 Measuring innovation using surveys

Although the Oslo Manual definition of narrow innovation is far from perfect, it is the basis of Community Innovation Surveys (CIS). They are a series of Innovation Surveys mainly undertaken every two years (or sometimes three yearly or annually) by National Statistical Offices in all the EU member states using a harmonised approach (with countries, where desired, making national additions). In some countries, participation in the surveys is compulsory; in others, it is voluntary.

Dive into the most recent Community Innovation Survey data.

In the UK, the CIS is called the UK Innovation Survey (UKIS). Participation is voluntary, but samples tend to be large. In 2021, the sample was nearly 32,000 UK enterprises with ten or more employees, with responses from around 14,000 businesses. In addition to questions relating to R&D activity, patenting activity, barriers to innovation and sources of information about innovations, the survey asks, crucially, whether firms are innovation-active, and whether the new products and processes are new to the firm, or the market, and first in the country, Europe, or the world.

UK Innovation Survey 2021 provides the results from the latest round of the UK survey.

innovation-active firm

A firm that has had innovation activities during the period under review, including those with ongoing and abandoned activities. In other words, firms that have had innovation activities during the period under review, regardless of whether the activity resulted in the implementation of an innovation.

The 2021 results for the UK suggest that:

• Between 2018 and 2020, 45% of UK businesses were innovation-active firms, with large organisations more likely to be innovative. That is a significant increase compared to 38% in 2016 to 2018. In 2018–2020, 58% of large businesses were innovation active, compared to 44% of small and medium enterprises (SMEs).
• In 2018–2020, 46% of businesses were broader innovators and 34% were wider innovators. Businesses were more likely to introduce new products than new processes. In 2018–20, 20% of businesses were product innovators and 16% were process innovators.
• Figure 13.1 shows a noticeable impact from the financial crisis upon all types of innovation types as well as a later decline since the period 2012 to 2014, with wider innovation falling by eight percentage points. The largest apparent fall was in the percentage of businesses introducing new business practices and new marketing methods. This decline may partly be an artefact resulting from an increase in the number of wider innovation options being added to the surveys covering the period 2016 to 2020 compared with the survey covering the period 2014 to 2016.

In these surveys, information on firm size is usually collected, but not other firm characteristics. In some countries, researchers may be given privileged secure access to anonymised company-level data so that CIS data can be combined with Census of Production data. When this happens, we can analyse the relationship between innovation, firm characteristics and innovation performance.

For cross referencing in the UK, it is possible to merge the information from the UKIS with the information contained in micro datasets on turnover, purchases, employment costs, capital expenditure and stocks.

One finding from the CIS data is that firms that undertake one innovative activity also tend to undertake others. Two major clusters of innovations tend to be undertaken jointly:

• Marketing, organisation, management and strategic innovations. This is labelled in the CIS data, somewhat confusingly, wide innovation.
• Machinery, process and product innovations. Labelled as technological innovations.

The evidence also indicates that the sample of firms can be separated into three clusters.

The smallest cluster undertakes both sets extensively, and the second (a slightly larger group), undertakes both to a limited degree, whereas the largest group does not innovate in either category to any degree.

This suggests not only complementarity, but also that wide and technological innovation are not substitutes, and, most importantly, that businesses that are innovative in one dimension tend to be innovative in all dimensions. There are agile organisations that realign not only to deliver innovation but importantly, extract value from it. The data patterns also indicate that the innovative performance of firms is heavily skewed, with a large part of the UK companies engaging only marginally in the innovation process.

ONS Resource

A further source of data on the uptake of management practices in British production and services industries is the recent Management and Expectations Survey (MES) carried out in 2020 and developed as a follow up to the 2016 pilot Management Practice Survey (MPS).

The MES questionnaire covers a slightly modified and broader set of questions than the MPS and the UKIS, and it was designed to align to the Management and Organisational Practice Survey (MOPS) conducted by the US Census Bureau.

Despite being one of the few surveys on the adoption of management practices, it contains limited information on either narrow or wide innovations and general company characteristics.

Outside the EU, there is survey data for many other countries. This data is collected and harmonised by international organisations.

The United Nations Educational, Scientific and Cultural Organization (UNESCO) Institute for Statistics provides free access to internationally comparable data on science, technology and innovation. It is the only global database of cross-nationally comparable indicators on innovation.

The latest report, compiled in 2015, examined the innovation process of countries in two groups of World Bank classification of level of income: high-income, and low- and middle-income. The survey gathered data from 71 countries, including nine in Africa, in order to produce a range of basic innovation indicators to inform policies at national and international levels.

The 2015 UNESCO main results can be found at Results of the 2015 UIS Innovation Data Collection.

Highlights from the 2015 survey:

• The most frequently implemented innovations are marketing methods.
• The acquisition of machinery and software is a frequently performed innovation activity.
• There are relatively few connections between manufacturing firms and higher education institutions despite the contribution these institutions make to knowledge creation and dissemination.
• Limited financial resources within firms are the main factor hindering innovation.

In high-income countries:

• Marketing was the type of innovation most frequently implemented by firms.
• The share of large innovative firms is above 50% in 30 out of 35 countries for which this breakdown is available.
• The shares of large innovative firms vary from 8.3% in Hong Kong Special Administrative Region of China to 92% in Luxembourg.

In low- and middle-income countries:

• Process and organisational innovations were most frequent among firms.
• The share of large innovative firms is above 50% in 11 out of 18 countries for which data are available.
• The shares of large innovative firms vary from 20.6% in South Africa to 89.7% in Costa Rica.

13.3.3 Using product launch data

Although CIS data are comprehensive, they are problematic for three reasons:

• Innovating firms, not innovations. They relate to the number of enterprises introducing innovations in different periods, but do not provide any insight into the number of innovations introduced by those firms.
• Local, not global. CIS-type surveys capture only product innovation by firms that produce in the sample countries. They do not include product innovations that are imported.
• No startups. Product innovations by newly established firms or entrants are not covered by the survey sample.

A particularly informative set of data is provided by Paul Stoneman’s book on launches of new creative products such as books and recorded music. For example, in 2006, 33,524 new music albums were launched. In publishing it was estimated that over 200,000 new titles were published in the UK in 2005 (growth of about 12.5% in the stock).

Stoneman P (2010), Soft Innovation: Economics, Product Aesthetics, and the Creative Industries, Oxford: Oxford University Press

Product launch information would seem particularly useful to address these problems. These include product launch data from government sources such as the Food and Drug Administration in the US and similar agencies elsewhere in the world, and product launch data from organisations such as those included in the Mintel New Products Database.

13.3.4 Defining successful innovation

In section 13.3.1 we argued that innovation might only be considered significant if it was a market success. For example, of the 33,524 new music albums launched in the UK in 2006, only 233 entered the top 40, and a US study estimated that despite huge numbers of new books being published, 205 books accounted for 84% of sales in a six-month period (and thus most books sold very few copies). This suggests there is much more innovation than successful innovation.

Recall the definition of innovation used by the Innovation Unit of the Department for Business, Innovation and Skills in the UK. It differs fundamentally from the Oslo Manual definition because it required the successful exploitation of new ideas.

The emphasis on success stresses the outputs, which the Oslo Manual definition does not mention. It emphasises that investment in innovation activities is of little economic relevance unless it successfully generates value. It also brings one closer to more recent definitions of innovations that are designed to reflect the contributions of investment in knowledge (or intangible) capital. In other words, innovations add to that capital only if they are successful.

How it’s done Measuring successful innovation

Giuliana Battisti and Paul Stoneman formalised this definition as the contribution of innovative activity to firm profit growth. This aims to capture the successful exploitation of new ideas.

This contribution is calculated using publicly available accounting data, for an unbalanced sample of 16,457 firms over the period 1988 to 2012 operating in 39 sectors and in 38 countries.

They found that, on average, the innovative activity is responsible for 5.15 percentage points of the company’s annual profit growth. The contribution of innovativeness to profit growth shows a very high variance across firms with significant differences within sectors, or countries, greater than differences across sectors, or countries.

You can investigate their methodology in their paper:

Battisti G, Stoneman P (2019), “Measuring the Innovativeness of Firms”, ESCOE Discussion paper No 19/2019, London, November

This approach has many similarities with the extensive literature on the growth of Multi Factor Productivity (MFP), as discussed in Chapter 6, which is often interpreted as a measure of innovative activity.

Total factor productivity measures how much output is produced per unit of input, including labour hours, the quality of labour and the impact of capital. It is the unexplained residual after the impact of these other factors is identified. It can be considered the “recipe” or the way in which the inputs are put together to make the output. So it is defined and calculated as the rate of growth of real value added that is not the result of increased capital and labour inputs.

Discovering a better recipe means more output for the same inputs and therefore higher MFP. Clearly this may be interpreted as successful innovation.

13.4 Measuring wider innovation using research and development

research and development (R&D)

Creative and systematic work undertaken in order to increase the stock of knowledge – including knowledge of humankind, culture and society – and to devise new applications of available knowledge. Data relating to research and development (R&D) has been collected for the longest period of time and very widely. It is now generally available for most economies, in some cases extending back to the 1960s, at both macro and micro levels.

The wider innovation concept encompasses all three stages of the Schumpeter trilogy: invention, innovation and diffusion. The first indicator is research and development (R&D) spending.

13.4.1 Measuring research and development

ONS Resource

In most countries, R&D data are collected by national statistical offices either annually or through innovation surveys. In the UK it is available at the ONS under Research and Development expenditure.

Data is also made available in comparative tables by several international organisations such as the OECD in its STAN database from 1987, or the World Bank in its Science and Technology databank, covering 264 countries.

As recommendations on accounting practice have changed over time, so the number of firms reporting R&D in their annual accounts has also been increasing, thereby making available an increasing amount of data at the firm level. Such data have been and sometimes are listed in scoreboards.

The 2022 edition of the EU Industrial R&D Investment Scoreboard covers the R&D activities of the 2,500 companies investing the largest sums in R&D in the world in 2021. In the past, the UK has also produced a local Scoreboard, but this was discontinued in 2010.

Guidelines on the collection of R&D data have been available from (and to a large degree followed by) most statistical agencies since 1963 when the OECD first published its guidelines. The latest edition of the guidelines says:

Research and experimental development (R&D) comprise creative and systematic work undertaken in order to increase the stock of knowledge – including knowledge of humankind culture and society – and to devise new applications of available knowledge.

Note that the definition covers both science and technology, and by including the phrase “knowledge of humankind, culture and society”, also includes the humanities and social sciences. The definition encompasses basic research (research with no particular application in view), applied research (research with a particular applied objective) and experimental development.

The Frascati Manual is a document that defines how we collect statistics about R&D. The Manual was prepared and published by the OECD.

13.4.2 GERD, BERD and other measures of R&D

gross expenditure on R&D (GERD)

Monetary value estimates of research and development performed and funded by business enterprise, higher education, government, UK Research and Innovation, and private non-profit organisations.

Data available usually includes total expenditure on R&D, labelled gross expenditure on R&D (GERD), which is often separated between civil and military. This expenditure is commonly also separated out between basic research, applied research and experimental development, at least in official statistics.

A further distinction generally made is that between the sectors funding R&D and the sectors performing R&D.

There are three main funding sources: business enterprises, government, and overseas sources (including international bodies such as the European Commission). In the UK, the main performers of R&D are business enterprises, government, including government departments and government-owned research establishments, and universities.

business enterprise R&D (BERD)

The component of GERD incurred by units belonging to the business enterprise sector.

The R&D performed and funded by business enterprises are usually given the acronym BERD, standing for business enterprise R&D data. It is collected regularly by statistical offices. Data is also collected on R&D performed by higher education (HERD), government, including UK Research and Innovation, R&D (GovERD), and private non-profit organisations (PNPRD). Business enterprises generally undertake more R&D than they fund, whereas governments fund more than they undertake. Figure 13.2 illustrates the different flows of research and development funding and performance in the UK in 2018.

We can use this data to compare R&D over time and across countries, but we must take care to use an appropriate price deflator. General price deflators such as the GDP deflator or the consumer price index would be unlikely to be appropriate.

In the UK, the Office for National Statistics (ONS) constructs a price deflator by weighting whole-economy materials and services input (producer) price indices together with a bespoke R&D labour cost price index. This labour cost price index is calculated using information on the labour employed in producing R&D coupled with gross hourly wage estimates for relevant occupations taken from the Annual Survey of Hours and Earnings to estimate a Laspeyres fixed-weight wage index.

International comparisons of effective R&D spends will also necessitate some correction for differences in the currency in which they are measured. The use of purchasing power parity is common and sensible.

In Figure 13.3, we illustrate some of the data on international patterns of R&D spending as a percentage of GDP in six economies.

We simplified this figure from a dataset of 41 countries that you can download from Eurostat.

The international patterns of R&D spending, as a percentage of GDP (measured in the complete Eurostat dataset) include the following:

• In 2019, the percentage of GDP spent on R&D fluctuated between 0.19% and 4.63%.
• Among the countries that invest most in R&D relative to GDP are South Korea, Sweden, Japan, Germany and the US.
• Italy and Spain typically spent a low proportion of GDP on R&D relative to other European countries.
• Russia and the Eastern European countries such as Romania and Bulgaria again spend less.
• R&D expenditures in the EU have increased slightly over time, averaging 2.22% in 2019 in comparison with 1.78% in 2005. However, substantial differences remained within the EU, with only Sweden, Germany, Denmark and Austria spending 3% or above.

13.4.3 The limitations of R&D as an indicator

Despite the popularity and wide availability of R&D data, this indicator has a number of limitations as a measure of innovation, meaning we might underestimate total innovative activity and provide an unbalanced picture of overall innovative activity:

• Not novel, still innovative. The Frascati Manual requires “an appreciable element of novelty or resolution of scientific and/or technological uncertainty”. But many new products placed on the market will not involve such novelty or resolution, especially if one is considering innovation in the creative industries (new music, video games or books) or the financial sector (new financial instruments).
• Design. This means that the source of many new products is not counted. In the same vein, expenditure on activities that have an aesthetic rather than a technological or scientific component, and which might well include product differentiation activities and marketing innovation, involving activities such as design, are also not defined as R&D.
• Most of R&D expenditure is staffing costs. In many small and medium-sized companies, due to limited resources, researchers and inventors responsible for innovations tend to cover more than one role within the organisation. They are then not classified as R&D staff and are not measured in official statistics.
• Process, not outputs. If the efficiency of R&D differs across time, countries or industrial units, then comparisons of R&D will not be a good measure of the amount of innovation produced. In fact, R&D spending may only be one input into innovation activity and the outputs produced could well be related to the other inputs involved, for example, prior knowledge. R&D alone may then be a poor indicator of outcomes to be expected.

13.5 Measuring wider innovation using intellectual property rights

intellectual property rights (IPR)

The general term for the assignment of property rights through patents, copyrights and trademarks. These property rights allow the holder to exercise a monopoly on the use of the item for a specified period. These cover inventions, literary and artistic works, designs, and symbols, names and images used in commerce. They are protected by the law and provide exclusive rights to the inventors to earn recognition or financial benefit for a certain period of time.

The second indicator of the rate of “wide” innovation are counts of intellectual property rights (IPR) granted.

13.5.1 Types of IPR registration

Court enforceable IPR are formal methods designed to protect the rights of the originator, inventor, or pioneer of a technical invention (which is both new and inventive when the patent application is filed) by discouraging imitation.

They also provide a monopoly right to use the invention for economic and commercial advantage, allowing the inventor to recover the often high research and development costs sustained. By guaranteeing that invention is the property of the inventor, the latter is able to capture the economic value of their inventions and this creates the necessary incentive and environment for further innovations. IPR are legally enforceable, although the extent of their enforcement and protection may vary from country to country.²

The four main institutional formal arrangements that exist in most countries in order to reinforce IPR are patent, design right, copyright and trademark.

Patents

patent

A right granted by a government to an inventor in exchange for the publication of the invention; it entitles the inventor to prevent any third party from using the invention in any way, for an agreed period.

Patents protect new inventions and cover how things work, what they do, how they do it, what they are made of, and how they are made. A patent gives the owner the exclusive right to stop others from making, using or selling their invention. In many countries, only advances of an industrial nature can be patented, which causes them to be labelled utility patents. (In some other countries, such as the United States, there is also a form of patents called design patents that correspond to the design rights that offer protection elsewhere.)

To be granted a utility patent, an invention must be new, must involve an inventive step that is not obvious to someone with knowledge and experience in the subject, and must be capable of being made or used in some kind of industry. Patent counts are sometimes used to measure R&D productivity, but this has limitations as a measure. There will be many innovations that fall outside these restrictions and therefore are not patentable. Once granted, a patent must (in the UK) be renewed every year after the fifth and may then provide protection for up to twenty years from the date of the patent application.

Design rights

design rights

Granted by different jurisdictions as a form of formal intellectual property protecting ornamental properties, shapes, configurations, appearance, or pattern of an article of manufacture, and such designs must be both new and distinctive or original. Can be referred to as “design patents”, “community designs” or “registered designs”.

Design rights apply to intellectual property but relate to the physical appearance and shape of a product rather than the function or operation of that product, emphasising instead the appearance that results from the features of physical products or the way they look. They may, therefore, offer protection to product innovations that would not be covered by patents, for a fixed period up to 25 years; or they may provide additional protection to the design aspects of a new product over and above that offered by patents.

To qualify, the design must be new and individual in character; less than twelve months since it was first publicly disclosed; not dictated only by how the product works; and not include parts of complicated products that cannot be seen in normal use (for example, vehicle engine spare parts, or the parts inside a computer).

Copyright

copyright

A form of intellectual property rights (IPR) that grants the creator of an original work (creative work) certain rights over that work for a limited period of time. The copyright holder has an exclusive right to reproduce their work in various forms such as printed publications or sound recordings.

Copyright applies to the expression of an idea, not to the idea itself or to any process by which that idea is embodied in a physical artefact. This is in contrast to the patent system, where the idea itself is protected and owned for a period by the patent holder. But copyright does not protect the names, designs, or functions of the items themselves.

Copyright is particularly applicable to new products in the creative industries and protects sound recordings, films, broadcasts, and original artistic, musical, dramatic, and literary works, including for example photographs, art sculptures, websites and web content, computer programs, plays, books, videos, databases, maps, and logos. For example, copyright is the mechanism that may be used to protect originators from the unauthorised downloading of music and films from the internet, or against copying from others’ originals. Software enjoys only copyright protection and not patent protection. In the UK, copyright is an automatic right and does not need to be applied for – though this is not true in all countries. This means it would be very difficult to discover the number of copyrights claimed.

Trademarks and trade secrets

trademark

Refers to words, symbols or other marks which are used by firms to distinguish their products or services from those offered by others. They protect inventions such as machines and machine parts; the design, shape or look of a product; things that are written, made or produced; and the name of a product, the logo or the brand.

A trademark is a sign that distinguishes a firm’s goods and services from those of other traders. A sign includes, for example, product or service names, logos, jingles and more generally words, pictures, or a combination of these.

Whereas patents require novelty and copyright requires originality, the counterpart for trademark is distinctiveness. While patents are not available for aesthetic innovations, such innovations may be trademarked.

Non-aesthetic innovations may also be trademarked. For example, a rock group can trademark its name, a product with a particular aesthetic, such as the iPad, can be trademarked, and particular products such as Mars bars and Crunchie bars may also be trademarked. To be registerable, the trademark must be distinctive for the goods and services for which application is made and must not be the same as, or similar to, any earlier marks on the register for the same or similar goods or services.

In the UK, a trademark does not have to be registered. An unregistered trademark provides certain rights under common law and the owner can use the trademark ™ symbol. However, it is easier to enforce rights if the mark is registered, which also enables use of the ® symbol to indicate registration.

13.5.2 Counting IPR

Counts of registered, hence formally recognised, IPR are an obvious measure of the extent of innovation. Of the various protections available, copyrights are not easily monitored, but patents, design rights and trademark registration data are more reliable and easily accessible. Hence, we focus on the latter group and report in Figure 13.4 the number of applications and registration at the UK Intellectual Property Office (IPO) from 2012 to 2021.

Both the numbers of patent applications and grants have declined significantly since 2012. On the other hand, trademarks and designs have shown remarkably sustained growth, suggesting a clear shift in the nature of the typology of ideas being generated. Whilst the rapid change from 2020 to 2021 is in large part due to the clearing backlogs arising from the coronavirus pandemic, it doesn’t take away from the clear upwards trend present in the previous years.

In the UK, the IPO holds databases that allow users to search by keyword, phrase, image or reference number. Existing patents, trademarks and designs can be identified, as can decisions on previous trademark applications. This is because access to searchable datasets of IPR is not just interesting for national statisticians. Inventors use them too:

• Is it protected? They must check whether their ideas, or their elements, have already been implemented by someone else.
• Does it have commercial application? They want to identify business opportunities, potential sources of collaboration and exploitation through licenses and formal agreements.
• Can I solve a problem? Technical holdups can be solved by licensing someone else’s innovation.
• What are existing patents worth? A count of citations in the various technological fields show this.
• What are trends in technology? They can explore what competitors are doing.

The same information is important, too, for investors, government, intergovernmental organisations and for academic research. By exploring available details on individual patents registered, one can obtain very fine micro data sets such as the number of applications and grants by country of origin, patent sectoral classification and name of applicant.

How it’s done Protecting intellectual property with patents

In 2022, for the tenth straight year, Apple Inc. was ranked the world’s most valuable brand (worth $482 billion) by Inter Brand, a business that compiles this statistic every year.

To protect its inventions and to deter competitors and imitators, the company has made extensive use of formal and informal intellectual property rights. Apple has consistently enforced secrecy until the launch of products and used complexity as a way to deter lookalike attempts.

The value of Apple’s products resides in both the function (protected by patents) and aesthetic design appearance (protected by registered designs, referred to as design patents in the US) to protect the latter.

In 2011, Apple started to claim its design patent over Samsung and began litigating against the South Korean company in technological patent infringement suits in a US court.

At the core of the case were three registered designs. The first registered design focused on the front flat clear face, the second on the bezel (the borders between a screen and a phone’s frame) and the third on the “ornamental design for a graphical user interface for a display screen or portion thereof”.

Apple argued that several of Samsung’s Android phones had systematically copied Apple’s innovative technology and products, features and designs, as prior to the introduction of the first generation iPhone (launched in 2007), there were no cellular phones that used a rectangular display screen that allowed for touch control interface.

The case was expanded to 50 lawsuits. The patent war between the two companies lasted more than seven years. In 2012, the American jury ruled in favour of Apple and ordered Samsung to pay Apple about $1.05 billion, the most significant design patent infringement award of all time.

By including design patents into its intellectual property portfolio, Apple was able to protect and capture the commercial exploitation of its innovation based upon its aesthetic appeal (rectangular shape multi-touch) rather than its functionality. However, a series of appeals followed up until 2018 when the order of dismissal of the first Apple v. Samsung case in the Northern District Court of California declared that the litigation was over, with both sides reaching an undisclosed out-of-court settlement.

13.5.3 IPR datasets

There is a wealth of data available on IPR, which can easily be searched through IP Offices’ ad hoc search engines and is freely accessible on their websites. For example:

• The European Patent Office (EPO) collects and provides data on European patents.
• The European Union Intellectual Property Office (EU IPO) provides data on EU trademark and registered Community design .
• The US Patent and Trademarks Office (USPTO) compiles an annual dataset that contains counts of in-force and issued patents from 1840 by industrial sub-category and data on trademark applications filed with or registrations issued by the USPTO since 1870.

International datasets have also been compiled, collating the information across the various national IP offices:

• The EPO produces and maintains the Worldwide Statistical Patent Database (PATSTAT) which is extensively used for statistical analysis. It is based on bibliographical and legal event patent data collated from leading industrialised and developing countries relating to millions of patent applications of more than 80 countries.
• The EPO also provides ESPACENET, the largest single source of technical information in the world containing millions of patent documents alongside machine translation of patent documents into a range of languages and text search tools. Machine translation services are provided also by the Japan Patent Office (JPO), China National Intellectual Property Administration (CNIPA), the former State Intellectual Property Office (SIPO) of the People’s Republic of China, and the Korean Intellectual Property Rights Information Service (KIPRIS).
• The OECD produces the OECD Triadic Patent Families Database covering sets of patents filed for at the EPO, the Japan Patent Office (JPO) and granted by the United States Patent and Trademark Office (USPTO).
• A full-text search service of published international patent applications and a list of international patent databases is also provided by the World Intellectual Property Organization (WIPO) via PATENTSCOPE. WIPO collates and publishes worldwide data not only on patents but also on design rights, trademarks and copyrights, from 190 countries.
• The World Bank publishes the World Development Indicators giving longitudinal data on IPs spanning up to 59 years (from 1960 to 2019) for several developed and developing economies.

13.5.4 Using IPR to measure global innovation

Figure 13.5 illustrates a striking pattern in the number of patent applications across the world. Note the rapid increase of patent applications in China in the last decade in contrast to the relative decline in the US and Japan.

We could interpret this as the decline of innovation in two countries that dominated the adoption of process and technology improvement in the second half of the twentieth century, with China taking over. But the use of patent statistics as a measure of innovation has three problems:

• Not all innovations can be patented and not all patents lead to innovation. Patent statistics alone will give a distorted picture and need to be supplemented with statistics on, say, design rights and trademarks. Even then, in the absence of good copyright data, many innovations in the creative sectors will not be counted. Innovators may also try to limit emulation informally through secrecy, information-sharing arrangements, control of strategic resources, specialised knowledge, threats, and learning by doing or other such informal mechanisms. The statistics will not recognise such practices and their use may well differ by industry (defence sectors for example will more often rely on secrecy).
• Not all patents have the same purpose. Patents can be used strategically, and their economic and technological value can differ widely. For example, for years Japanese companies have emphasised the quantity of patents as a measure of R&D productivity, and this gives an incentive to reward R&D personnel, arguably leading to patent proliferation of mixed quality and value. Patents are also used strategically to strengthen protection through company bargaining power, deterrence and retaliation. Strategic patenting has clear implications for competition policy. The number of patents (shown in Figure 13.5) may be an imperfect indicator of innovation.
• Not all patents are of the same quality. Some will be more important than others, and thus a simple head count may be misleading. Attempts to correct for this have included using data on citations of patents by later applicants as a measure of quality (more citations means more importance) and also using renewals data, it being argued that more valuable patents will be renewed (although the size of the renewal fee will also matter).

Ariel Pakes and Margaret Simpson concluded that most patents turn out to be of little value, with rare big winners. They estimated a 1989 value of $6,000 for the average patent, with half of all the estimated patent rights accruing to between 5% and 10% of all patents. They also concluded that patents not renewed at or before five years had expired had almost no value.

Pakes A, Simpson M (1989), ‘Patent Renewal Data’, Brookings Papers on Economic Activity: Microeconomics, Volume 1989, pages 331–410

There is one major advantage of using patent data. IPR protection will often be sought by their owners for innovations being introduced into an economy from overseas. Importation is likely to be an important source of new products and processes and patent counts will reflect this more adequately than, for example, local R&D spend.

So IPR registration data has advantages and disadvantages as a measure of wider innovation activity. The data is widely available and there are long time series. But while R&D is clearly a measure of inputs into the innovation process, with IPR registration data the picture is less clear. Patents may be applied for at different stages in the innovation processes, sometimes early in that process, relative to the placing of an innovation upon the market, and sometimes late. Early application may be made, for example, to provide security for further development, while late application may provide greater secrecy.

Similarly, design rights may also be applied for at early or late stages, but since design will probably be an activity that is close to market, any application may be relatively late. By their nature, copyrights and trademarks will probably occur very close to or at the time of putting a product on the market.

Patent applications and design rights may best be considered as primarily a measure of intermediate outputs from the innovation process (those outputs reflecting increases in knowledge), with copyright and trademark counts being primarily a measure of innovations marketed rather than under development.

13.5.5 Informal methods of IPR protection rights

A further type of intellectual property rights is trade secrets (or confidentiality agreements) protected by the law via the enforcement of non-disclosure agreements (NDAs). NDAs are stipulated and signed by two parties who agree not to disclose to third parties the information agreed to be kept confidential. Employees, consultants and potential business partners may all sign NDAs that are not necessarily reported to or signed via institutional sources such as the IPO.

Should the information contained in the NDA be disclosed, it can be treated as breach of confidence and it is liable to legal action, although its protection can be weaker in countries such as the UK and the USA where trade secret law is part of common law.

There are some colourful examples of use of trade secrecy to protect IPR. One is the Coca-Cola secret formula whose only official written copy is allegedly kept in a bank vault in Atlanta. According to the legend (or possibly the marketing strategy) surrounding the trade secret, only two anonymous employees at any given time know the secret recipe and they are prohibited from taking the same plane in the event it crashes.

Another example is the trade secret of Colonel Harland Sanders’ secret blend of 11 herbs and spices used in Kentucky Fried Chicken (KFC). The recipe has been kept secret since July 1940 when Colonel Sanders, the founder of KFC, began selling fried chicken from his roadside restaurant in Corbin, Kentucky.

The use of informal intellectual property protection has been shown to be quite widespread and it is often preferred as the most effective protection mechanism for inventions. It is indeed cheaper, less time consuming and does not require complex applications.

From a statistical viewpoint, however, contrary to the formal protection methods, data on informal protection methods are neither systematically collected nor disclosed. Ad hoc surveys are the main source of information. Figure 13.6 reports the use of different approaches to formal and informal IP protection between 2018 and 2020 by the 13,598 enterprises present in the UK Innovation Survey (UKIS).

The results confirm the popularity of informal methods, although with differences across sectors. Knowledge intensive services tend to use more informal methods (lead time, complexity and secrecy) than high-tech manufacturing, while high-tech manufacturing uses more formal protection methods (trademarks and patents) than knowledge intensive services. It is worth pointing out that more than one type of informal and formal method can be used for a single product, indicating complementarities across protection methods.

13.6 Measuring wider innovation using diffusion statistics

Whether technological or non-technological, when an innovation appears upon a market for the first time, acceptance is often initially limited, with successful innovations then spreading more widely over time. This spreading process is known as innovation diffusion.

The third indicator of the rate of “wide” innovation is counts of innovation diffusion.

13.6.1 Measuring diffusion

The extent and speed of diffusion is a clear output measure of the innovativeness of an economy. In fact, of course, it is only as diffusion occurs that the benefits from innovation are realised. Diffusion can be analysed at country, sector, firm or at household level, depending on the nature of the innovation observed and the level of analysis. However, despite its importance, there are limited official statistics relating to the uptake of innovations. With few exceptions, ad hoc statistical surveys are the main sources of information.

Diffusion can take a very long time, sometimes decades, to spread across and within the population of eligible users.

To document the diffusion process fully, one needs to record, for each innovation, the number of users, and the depth of their use, from the date of first appearance of an innovation until diffusion is completed. Cross-sectional observations at a given point in time would provide only a partial view of the innovation diffusion process.

As an example of the need for a longitudinal view, the World Bank has published data on digital adoption across 180 developing and developed countries. Unfortunately, although interesting in showing the depth of penetration at a single point in time, it does not have the longitudinal dimension necessary to reflect the process of diffusion.

What data is available also tends to concentrate upon the number of users (the ‘inter’ dimension) rather than the depth of use (the ‘intra’ dimension). Diffusion data also has to be innovation specific, and it is often difficult to find agreement as to how to define the limits of any innovation being studied.

There are a few official statistics relating to technological and non-technological innovation diffusion:

• The pan-European dataset on information and communications technology (ICT) usage and e-commerce in enterprises is based on surveys conducted on behalf of the European Statistical Office (Eurostat) by the EU National Statistical Institutes or responsible government departments, using a common questionnaire.
• In the UK, the e-commerce and ICT activity survey has been carried out by the ONS across 7,700 businesses annually since 2000. It contains information on internet sales by businesses in the UK (total value and as a percentage of all retail sales); the percentage of businesses that have a website and broadband connection; and the use of Big data analysis, 3D printing and robotics.
• The UK also carries out a survey on ICT usage in households and by individuals on home internet, how we access it, for how long and what we use it for, including social media usage. It also includes the number of people who have used the internet those who use the internet frequently, and those who have never used the internet, broken down by age and gender.

ONS Resource

The ONS carried out the Management and Expectations Survey (MES) in partnership with the Economic Statistics Centre of Excellence (ESCoE) on the uptake of businesses’ management practices, as well as future expectations of turnover, employment, expenditure and capital investment growth. The survey covers 50,000 enterprises in the production and services industries in Great Britain. Data on the adoption of broader managerial innovations is also collected by the CIS survey.

None of the surveys asks for the date of first adoption of the innovation(s), and therefore one cannot derive the diffusion curves, but they do provide useful information of the depth and breadth of use across the population at each point in time.

For example, Figure 13.6 reports the diffusion of ICT in Great Britain measured by the percentage of households that used the internet between 1998 and 2020.

Typical of many diffusion processes, the diffusion pattern is one of a slow initial speed of diffusion that increases rapidly before gradually reaching the saturation point, when all the potential users have adopted the innovation. Also, characteristic of other diffusion examples (see Figure 13.9), it has taken nearly two decades for ICT to spread across the majority (93%) of eligible users.

ONS Resource

While Eurostat makes data publicly available only at sector, region and country level, ONS and the national statistical offices of other countries make them accessible at the finer business and household level, in an anonymised form, for academic and government research purposes under strict data management and non-disclosure conditions.

Anonymised data are made available through the ONS Secure Research Service and the UK Data Service (UKDS) and are available to any user, without charge, provided they have registered with the UKDS and they have undertaken the training to become an approved researcher.

13.6.2 Alternative sources of data on diffusion

There are a few additional sources to supplement the relatively limited data on diffusion. Government departments sometimes make available aggregate administrative data relevant to their competences. UK examples include:

• The Department for Transport in association with the Driver and Vehicle Licensing Agency (DVLA), publishes data on the number of licensed vehicles (and hence on their diffusion) over time by fuel type, for example, electrical, fuel cells, diesel, or petrol, from 2001 to 2018.
• Similarly, the Department for Business, Energy & Industrial Strategy produces various statistics, such as the Digest of UK Energy Statistics (DUKES), Energy Trends and regional statistics on the uptake of renewables such as solar panels and wind farms.

Trade bodies also publish data for their respective industries for particular industries:

• Data about robot usage is available from the Industrial Federation of Robotics.
• The Recording Industry Association of America provides time series data on the use of different recording media back to 1973, and the International Federation of Phonographic Industries collects similar data worldwide.

While trade body data is useful, it should be said that it is not always free to access. Further, while it is useful for identifying general aggregate trends at geographical, sector and national level, confidentiality considerations mean that such sources provide less detail for the microeconomic analysis of technology diffusion at household or firm level.

The Historical Cross Country Technology Adoption (HCCTAD) data set covers nearly two and a half centuries, from 1788 and to 2001, but is limited to the leading 23 industrialised countries in the world. CHAT is a data set with information on the adoption of over 100 technologies in more than 150 countries since 1800. The technologies that are covered relate primarily to durable goods, but include both consumer and producer durables, for example, mobile phones, TV sets, different steel production technologies, steamships, and motor ships. Such data is informative in comparing diffusion of single technologies across countries (see Figure 13.8 for an example) or to compare diffusion of various technologies within a single country (see Figure 13.9).

The diffusion curves for mobile phones in Figure 13.8 show the earlier adoption (date) and the faster speed of diffusion (steepness of the curve) of telephones uptake in the US. The adoption lag of the UK is interesting. Not much can be said about the height of the curves as they depend on the population size and hence the number of potential users. The population size is not accounted for in this exercise.

13.6.3 Patterns of diffusion

Paul Stoneman summarises evidence both on the extensive margin (the spreading across countries) and the intensive margin (spreading with countries) in this book:

Stoneman P (2002), The Economics of Technological Diffusion, Oxford: Blackwell

The figure of 47 years for adoption lag comes from the Comin and Hobijn study of 166 countries over the past two centuries. But there is wide variation across countries and across technologies.

adoption lag

The rate of pace at which new technologies spread and are exploited, with a lag referring to the length of time technologies remain unexploited. The adoption lag is therefore the difference in time taken to exploit a new technology after its release.

extensive margin

The extent to which innovation increases the number of products produced by each firm.

intensive margin

The extent to which innovation increases the value of each product produced.

For the majority of the innovations, diffusion curves are S-shaped while there are large cross-country and other differences in the adoption process. Figure 13.9 shows the diffusion of innovations in the UK in the last century. The rate of adoption can be affected by:

• Adoption lag. This is the time taken from invention to first adoption. In the case of international technology adoption, the adoption lag is generally long, on average 47 years, according to one study.³ 54% of the variance in adoption lags was explained by variation across technologies, 18% by cross-country variation, and 11% by the covariance between the two. The study also found that acceleration in technology adoption took place during the whole of the two centuries covered by the data, thus preceding the digital revolution or the post-war globalisation process.
• Faster spread between countries. The HCCTAD data set also suggests that the spread of a new technology to more and more countries (the movement of the extensive margin) is completed long before the spread of the extent of usage in each country (the movement of the intensive margin) comes to an end. Thus, although in the early stages of diffusion, movements in both the extensive and the intensive margin contribute to the advance of diffusion, in the later stages only movements in the intensive margin matter.
• Number of users or intensity of use? In a single country, there are two margins of interest. The first is the number of users (perhaps as a percentage of the population), inter-firm or household diffusion. The second is the intensity of use by each user, in other words intra-firm or household diffusion. Analysis indicates that the depth of use of new technologies lags well behind the spread of adoption across users (whether firms or households). This means that being a user does not necessarily imply being an intensive user of an innovation.
• Production and consumption or use do not coincide. To give an obvious example, the use of motor vehicles is different from their manufacture. But diffusion of innovation is relevant to both. So for many products, there are two diffusion processes that happen simultaneously, each with its own extensive margin. International trade is one mechanism in which production in one part of the world can translate into consumption elsewhere. Changes over time in the international patterns of production may therefore also be considered as part of the study of diffusion. Such issues have been addressed in so-called “North–South models” which have emphasised that the developed economies (the North) account for most of the worldwide innovations but that, over time, technological knowledge is diffused to other developing economies (the South), thereby spreading the benefits of innovation worldwide. As this spreading occurs, the manufacture of innovations should be moving from the developed to the developing countries. Data on such phenomena can be established by use of output statistics produced by most national statistical agencies.

The Battisti and Stoneman study of UK metalworking and engineering firms shows that inter-firm diffusion (measured via the number of users) is dominant during the early stages of diffusion, while intra-firm diffusion (measured via the share of output produced using the new technology) becomes the main driving force in the last stages of the diffusion process. The findings also indicated that the number of users was an incomplete indicator of the overall usage of a new technology.

Battisti G, Stoneman P (2003), ‘Inter- and Intra-firm Effects in the Diffusion of New Process Technology’, Research Policy, Volume 32, Issue 9

A useful review of North-South models is given by:

Afonso O (2011), ‘R&D Direction and North-South Diffusion, Human Capital Growth and Wages’, Economics Research International, Review Article ID 401928

Using television as an example, Zhicun Gao and Clem Tisdell have explored patterns of the worldwide production of both black and white and colour TV sets. They show production shifting over time from the UK and US to Europe and Japan, and then to Taiwan, South Korea, Mexico, Italy, and Malaysia, all relatively low-income countries, and to China.

Gao Z, Tisdell C (2004), Television Production: Its Changing Location, the Product Cycle and China. Working Paper 26, Brisbane School of Economics, University of Queensland

13.7 Measuring innovation using alternative indicators

13.7.1 Prices

Surprisingly little use of pricing data has been used to illustrate where prices are falling over time because of innovation. Conceptually, allowing for quality change in the measurement of price change is a long-established practice (see Chapter 1) at aggregate level and there are well-established techniques to calculate such adjustments, although their use is not widespread. The same techniques can be used to measure the effects of quality change on the prices of individual products. Construction of hedonic price indices, for example, indicates how product prices have fallen as the result of product innovation improving quality per unit price. This has been done widely for personal computers, laptops, digital cameras, and other electronic goods, but is less common to date in other areas.

13.7.2 Scoreboards

A different approach is to consider jointly many indicators in what are known as scoreboards. They are widely used by government and international organisations for benchmarking innovation performance, and to identify areas of strengths and weaknesses for policy intervention. Examples include:

• The OECD Science, Technology and Industry Scoreboard uses indicators that include not only traditional R&D employment, spending and funding data, and various patenting data, but also include data on broadband penetration, M2M and SIM Card penetration, internet usage, the emergence of ICT technologies, citation of scientific publications, robotisation, labour and multifactor productivity, ICT footprint, and venture capital investment.
• The European Commission’s European Innovation Scoreboardincludes some 27 different indicators deemed to be relevant to innovation for EU and non-EU countries including the UK.

Scoreboards of this kind have the advantage of presenting a range of information in a convenient form. The difficulty is to know how much weight to give to the individual indicators, which may well be moving in diverse directions, to draw out the overall story. The European Commission Scoreboard attempts to give some help here by calculating a Summary Innovation Index for each Member State, giving equal weight to each of the available indicators. But such a weighting scheme is necessarily arbitrary.

13.8 Conclusion

Innovation is the engine of growth. We have identified wide (new to the individual economic unit) and narrow (new to the country) concepts of innovation and also differentiated between the process of innovation and innovation as an output of that process.

One of the significant problems in measuring innovation is that it requires an understanding of what needs to be measured and awareness of what can be reliably measured. Issues arise in this area when measuring business innovation, as more quantitative measures of innovation activities are more difficult to capture. Innovation and digitalisation are playing an increasingly important role, emphasising the need to create a rich, varied dataset that can be compared internationally.

There exists a plethora of micro and macro data on innovation accessible via national statistical offices, international organisations, intellectual property offices, firm level accounts and diverse other sources. Such data are widely used not only to measure the innovative effort and output of companies and countries worldwide, but also to explore the emergence of new technologies. Historically, such data have been one of the main inputs driving government innovation policies.

It is clear however that although there is wide availability of data on the level of activity in the innovation process (such as R&D data and data on IPR), there is much less data on the output from the innovation process such as the extent of the uptake of technological innovations by final users (either households or companies) or the depth of use by the adopters. Most importantly, there is less data on the benefits of uptake and depth of use, as it is only when ideas and innovations are used that the benefits from innovation are realised. Although Innovation Surveys have yielded much on this issue, these surveys have their limits and could be further extended.

13.9 Further reading

• Afonso O (2011), ‘R&D Direction and North-South Diffusion, Human Capital, Growth, and Wages’, Economics Research International, Review Article ID 401928
• Battisti G, Stoneman (2019), ‘Measuring the Innovativeness of Firms’, ESCOE Discussion paper No. 19/2019
• Battisti G, Stoneman P (2003), ‘Inter- and Intra-firm Effects in the Diffusion of New Process Technology’, Research Policy, Volume 32, Issue 9
• Baumol, WJ (2002), The Free Market Innovation Machine: Analyzing the Growth Miracle of Capitalism Princeton, NJ: Princeton University Press
• Comin DA, Hobijn B (2010), ‘An Exploration of Technology Diffusion’, American Economic Review, Volume 100, Issue 5
• Gao Z, Tisdell C (2004), ‘Television Production: Its Changing Location, the Product Cycle and China’, Working Paper 26, Brisbane School of Economics, University of Queensland
• Office for National Statistics (2018), ‘Gross Domestic Expenditure on Research and Development, UK: 2018’
• Office for National Statistics (2018), ‘Management Practices and Productivity in British Production and Services Industries: Initial Results from the Management and Expectations Survey: 2016’
• Organisation for Economic Cooperation and Development (2005), ‘Oslo Manual: Guidelines for Collecting and Interpreting Innovation Data, 3rd Edition’
• Organisation for Economic Cooperation and Development (2015), ‘Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development’
• Pakes A, Simpson M (1989), ‘Patent Renewal Data’, Brookings Papers on Economic Activity: Microeconomics, Volume 1989, pages 331–410
• Schumpeter JA (1934), The Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest and the Business Cycle, New Brunswick, NJ: Transaction
• Stoneman P (2002), The Economics of Technological Diffusion, Oxford: Blackwell
• Stoneman P (2010), Soft Innovation: Economics, Product Aesthetics, and the Creative Industries, Oxford: Oxford University Press
• United Nations Educational, Scientific and Cultural Organization (2017) ‘Summary Report of the 2015 UIS Innovation Data Collection’

Notes