When we think innovation - is it always about more or better

You are probably familiar with the 5 types of innovation:

1. new products
2. new processes
3. new supplies
4. new markets
5. new forms or organisation (typically at the firm level)

Can we think about innovation at the entire system organisation level - innovating in the ways we do business?

Take the following TED video that highlights massive food waste. The point is that a sustainable approach to our global population challenges is to simply waste less food. To do this we need to retro-innovation and make the food standards in supermarkets less perfect. Not that food quality will suffer just the cosmetic appearance of some of our food.

Macro-Innovation, and I-O II (No.3) Industry models

Just as a quick follow up to the last two entries, I caught this last paragraph in a paper published in just the last couple of weeks. This quote very nicely sums up the argument of the last 2 blogs.


Related variety may be defined in terms of membership of two-digit national industry statistical categories, such as “chemicals” and, in more modern classifications, “life sciences” (Boschma, 2005). However, from an evolutionary perspective, which is necessary to explain clusters and platforms, the "relatedness” may be visible only after the fact. An example today is the close interaction between agro-food and automotive research and business activity around biofuels. These interactions show their “relatedness” unpredictably and only ex post. (page 1423)

Nicely put Phil, industry models are not supply chains or even related variety. However, so long as we are willing to watch industry and forget certain cherished analytical approaches to industry analysis we can do ex-post and even some careful prospective analysis early on.

By the way, the quote comes from:


Philip Cooke (2012): 'From Clusters to Platform Policies in Regional Development', European Planning Studies, 20:8, 1415-1424.

Macro Innovation - I-O II

So in the last post I suggested that when we analysis innovation effects at the macro level we need to think differently.

At the recent AAAS meeting in Vancouver I surprised to learn the 'plastic cups' we were drinking from were in fact corn starch. I don't know which company they were from but here is a picture of what is possible:

http://www.flickr.com/photos/becs027/2318310270/
They look like plastic, they feel like plastic and they perform like plastic - except they are compostable.

So what difference would this make to an I-O table. Now I haven't looked into the exact processes of the technology so this is just suggestive at the moment.

(click on image to enlarge).

I have suggested here that the changes in the I-O table would be minimal. Some added agriculture into the plastics industry with some necessary increase in chemical / petroleum products into the ag industry to increase production. The petroleum industry's input into the plastics industry would decrease.

Okay, so to a different example; the digital camera revolution I discussed last time.
So just a note. As it happens - the Canadian I-O table which I am using as an example doesn't have the category 'C33 Medical, precision and optical instruments' populated with data - it is amalgamated with a different category. But that is okay for this example, I have simply highlighted the appropriate cells.

First we need to include not just the B2B but the entire sheet because the impact has been with final consumers.

Chemicals has dropped and with it the entire column of purchases, as well as supplies through to retailers and consumers. Optical devices have increased along with the entire column as more devices include optics.

Finally to some speculation. There has been much talk lately about 3D printing. Now here again this approach can help us think more clearly about the possible effects of such technology and why it become highly problematic to talk sensibly. The big benefit of 3D printing is that it cuts down on waste - additive manufacturing as it is called. That changes the coefficients.

Second it makes manufacturing possibly more local.

I would suggest that at the very least 3D printing at some point has the potential to change the very structure of manufacturing both in what it supplies and how it is supplied and in what it buys.

So in this way I would suggest we can use the structural data we already have but then develop new approaches to analysis that don't take existing structures for granted but begin to layer it with new meaning that represent changes in the operating models of segments of economic activitiy.

Conceptualising innovation at the macro landscape level

Recently I have been puzzling over the idea of whether we could look back across time and measuring the impact of a particular singular or group of innovations.

The problem is very rapidly you run up against numerous problems some of which are intertwined.

So lets start with a simple example; the transformation of camera from film to digital. Papers that I have seen on this topic divide on how they deal with it. One type of study is the case method that seamlessly describe both the camera manufacturers and the film manufacturers and suppliers such as Kodak and Fuji.

The second method often uses some case data but also draws upon industry data.

And here is the problem; film manufacturing is in the chemicals industry and photographic equipment has essentially its own category and never the twain shall meet. Standard industry data cannot address the interactions between these categories. If you you turn to I-O data that won't help much either. Input-output data helps us to see interactions where those interactions are a part of actual the transformation of products such as minerals into steel and steel into cars.

Film and Cameras have no such interactions.

So this is point 1 - there are common sense business models that link complementary products for which there is no obvious data.

We then turn to another problem. We have come to understand innovation has been one of five categories:

• product innovations
• process innovations
• new suppliers (inputs)
• new markets (outputs)
• new organisational structures

But if we look at innovation in the aggregate there is a conundrum. The example above is obviously product innovations isn't it? But what of the impacts:

• Product innovation - a camera transformed from mechanical to electronic.
• Process innovation - the entire manufacturing process for cameras has changed.
• Suppliers - camera companies needed to link up with electronics companies
• Markets - cameras are now in everything so the camera companies are selling to electronics companies. Retail market for film processing has collapsed. Film production has collapsed. The market for high quality lens making has probably also declined.
• All of this must have require organisational changes.

So point 2: from a post innovation aggregate point of view there are only innovations - dividing between categories is extremely difficult to determine. 

This begins to suggest that the innovation literature has an unexposed assumption nested within it. I won't call it a paradigm as that means something different. Such assumptions are extremely hard to reveal and map. I showed up two of these in my book Innovation Systems Frontiers. Innovation literature focuses on exports ignoring imports and does not ptoblematise political borders. But that took 7 years of literature mapping to highlight. I don't propose to do the same here.

So I will simply assert that scholars are more interested in corporate strategies for innovations and industry approaches to innovation. This developmental perspective includes all the geography literature. Now I realise that there is a growing research impacts literature but that is a different matter  which I want to leave aside in this blog.

So if you go here http://timkastelle.org/blog/2012/06/the-complete-innovation-matrix/ and read Tim's really interesting ideas on corporate innovation strategies you will see what I mean. I do encourage you to take the time and this blog is in part a response to think about the problem from the other end of the process.

We have spent less time thinking through the linkages in economies. We have wasted a lot of time in my opinion devising taxonomies (Pavitt) and technology flows (literature on embodied R&D), but the more this work rests on existing data structures rather than new surveys the more limited the results will be because of point 1 above.

A more idiosyncratic approach as an intermediate step may be a better starting point. If we adopt the model in Arthur's 'The nature of technology" we can construct examples such as that below. Lets start with one artefact - a car:

Endo-system technologies:

• engines
• glass
• electronics
• rubber
• plastics
• fabrics
• seats
• sensory systems (autopiloted cars)
• etc etc

Exo-system technologies

• fuels - processing quality
• fuels - sourcing geological science / deep sea drilling / oil sands etc
• roads
• traffic management
• reconfiguration of urban systems
• public transport systems 
• etc

So point 3: thinking backwards forces us to see that lying across our conventional data are multiple networks of intersecting business models and technological systems. 

What is my answer? Instead of looking for answers either within the existing data structures or ways to reform data collections, lets face the problem and working with the existing data build link activities that are currently not linked. Think of international airline network maps and you will have a vision of where I am going with this idea.

GERD to GDP and GERD / Gross International Product

Sometime ago I wrote in this blog suggesting that we need to be aware of the weaknesses of share of GDP as an indicator (whether it be for RandD or sectors) even though it is widely used in international studies.

There appear to be three primary circumstances where percentage of GDP is used as an indicator.

1. A through time measure of activity in a single country;
2. A cross country measure for a single point in time - this is an indicator of relative intensiveness; and
3. A cross country measure for a time period.

Of these three usages the first is completely legitimate, the second is mostly legitimate (although it should not be used alone) and the third I argue is highly misleading because of the differential growth rates of countries. #3 has some value but it is highly limited and needs to compared with a new indicator - which I have argued should be a panel GDP set. A country that grows rapidly will have a very hard time keeping RandD static let alone growing. Conversely, a country that has static GDP my find itself with constant or growing GERD?GDP.

The OECD amongst others continues to analyse a number of economic activities using the third approach. Here I want to highlight its use with RandD and show how much it changes the perspective if a different measure is created. For an example of the OECD use of the measure of RandD/GDP through time go here and watch the little movie.

In this blog I want to illustrate how I have gone about creating a panel dataset and how it can be used.

The criteria for a panel dataset are:
1. Long time series
2. High quality of data
3. Relevance (comparative GDP is log scaled so many small countries contribute little to the overall scale of a Panel GDP set)

Dating back as far as 1970, organisations such as the OECD have collected high quality GDP data for numerous countries. The number of countries to be included in the basket is somewhat arbitrary but for reasons of breadth of geographic coverage it seemed to me that the top 15 in 1970 provided a practical solution. This GDP G15 better reflects the advanced economy world of the last 40 years but lacks the BRIICS. The indicator could be named by various titles such as Benchmark International Product (BIP) or Gross International Benchmark Product (GIBP) but I think the simplest is Gross International Product (GIP)

The countries I have included are:

Australia

Belgium

Canada

France

Germany

Italy

Japan

Korea

Mexico

Netherlands

Spain

Switzerland

Turkey

UK

USA

GDP Dynamics

The first step is to understand GDP dynamics across a serious slice of time.
This first graph depicts in current $ PPP the pattern of GDP growth and relative scales (log).

Obviously, the USA is the largest economy with Japan second. But within this graph other patterns such as the rapid rise of Korea and Turkey are noticeable. The flattening out of Japan is also noticeable.

If this is converted to an image of proportions more interesting patterns emerge.

The USA is essentially steady in its share, while Japan and a number of the European countries loose share of GIP. Countries like Korea make up the difference and even Australia grows within the
panel.

GERD/GDP

So then we come to the classic RandD/GDP analysis. This is GERD data with Austria added in for GDP comparisons purposes.

It quickly becomes obvious that most countries have grown their GERD / GDP ratios. But then the question becomes how meaningful is this when we know that countries like Germany and Japan in particular have not been growing as fast as other countries in our panel set. For that analysis we turn to GERD/GIP

GERD/GIP

The image becomes very different when we account for different growth rates using our Panel set.

Now the GDP performance shines through. Japan through the 1990s had a falling share of GERD?GIP, Many European countries had fairly flat performance in the 1990s and 2000s

Though they are small, it is the small to medium economies - Austria, Canada, Australia, and Spain that do well.

So what you measures matters.



The advantages of a panel set (Gross International Product)

1. A panel set allows all countries to be compared on exactly the same basis.
2. the trend up and down are more persuasive because the growth or stagnation of individual economies is removed
3. it is possible to create groups of countries because the panel set is so much larger than any single economy.
4. Any number of countries can be analysed using the Gross International Product as the concept of 100% of GIP is essentially meaningless. If for example it was desired to look at BRIIC activity of a particular kind and together it exceeded GIP, that would not be a problem because of the way the measure works.



The Disadvantages
1. Scale becomes a more obvious factor - larger economies will look bigger on the graph. Thus, there is still some place for the tradition GDP measure but as is obvious it does not reveal the complete story.