Where Are Australia’s Most Innovative Postcodes?

According to IP Australia data covering live patents and applications, and recently-filed provisional applications, as at the end of 2015, the most innovative postcodes in Australia, across all fields of technology, are those taking in Macquarie University and surrounds in the North of Sydney.  Postcodes 1710 (Epping, NSW) and 2612 (represented by the nearby Blenheim Road post office) collectively accounted for nearly 1200 patents and pending applications, and 200 provisional applications filed by Australian applicants within the preceding three years.  This represents around 17% and 5%, respectively, of all such patents and applications owned by Australian applicants at that time.

Other hotspots are the regions around the CBDs of the New South Wales, Queensland and Victorian state capitals Sydney, Brisbane and Melbourne.  Also featuring strongly in the top 50 postcodes for patent filing are major university precincts in and around the University of Sydney, Monash University and the University of Melbourne.

In addition to the major cities, where it is unsurprising to find significant concentrations of innovative activities and industries, there are also a number of notable regional centres of innovation.  These include the NSW Central Coast, north of Sydney, particularly around Somersby (postcode 2250), west of Sydney around Long Point (postcode 2867), and Aitkenvale in the suburbs of Townsville in North Queensland (postcode 4814).

More generally, a map of Australian patent filing and ownership by postcode leads to a similar conclusion to the one I reached in my recent analysis of the origins of biotechnology patent applications – innovation is a collective activity that tends to occur in geographic clusters.  By exploring the interactive map below, you will be able to identify the various localities on national, state, regional, urban and suburban scale within which much of Australia’s innovative activity takes place.

The Map

Two types of information are represented in the map below.  First there is a ‘heat map’ layer, which indicates the intensity of patenting activity by entities within different geographic regions.  The more applications originating from within a region, the more ‘intense’ the colour – with red being more intense, or ‘hotter’, and green being less intense, or ‘cooler’.  The ‘dispersion’ of the heat, i.e. the extent to which it spreads out from the the sources of activity, scales with the zoom level.  Thus, as you focus in on particular areas, the ‘hot’ regions resolve into a more detailed representation of where patent filings have originated.

Secondly, the map includes markers, i.e. the blue circles, identifying the top 50 postcodes by patenting activity.  Clicking on any of these markers will cause a pop-up tooltip to appear, listing the postcode and name of the corresponding region, suburb or post office, along with its rank and the numbers of provisional and complete patents/applications originating from the postcode area.  Further details of how this information has been derived, along with a full list of the top 50 postcodes, are set out below the map.

Deriving the Data

The primary source of data for this analysis was the publicly-available Intellectual Property Government Open Data (IPGOD) 2016 data set, which I have used in a number of other recent articles.  In order to measure origins of innovation as at the end of the period covered by this data, i.e. 31 December 2015, I extracted:

1. all ‘complete’ (i.e. standard and innovation) patents and applications naming at least one Australian resident applicant and still in-force at the time; and
2. all provisional applications filed within the preceding three years, i.e. on or after 1 January 2013.

This selection resulted in a set of 34,999 patents and applications, collectively owned (either individually or jointly) by 18,202 Australian resident applicants.  I have assumed that all of these applicants may be treated as ‘active users’ of the patent system, since they have filed an application of some description within the preceding three years and/or are continuing to prosecute one or more applications, or to maintain one or more patents, that they have previously filed.

In determining rankings, and weighting of ‘heat sources’ on the map, I allocated one point for each complete application, and one quarter point to each provisional application, on the basis that, overall, Australian applicants file about four times as many provisional as complete applications.  Broadly speaking, this implies that only around one in four provisional applications lead to subsequent filings with the potential to result in granted patent rights.  It therefore seems reasonable to ‘devalue’ provisional applications accordingly.

I have used a number of data sources to associate each Australian applicant with a corresponding postcode.  Again, the primary source is the IPGOD 2016 data set, which includes geolocation information (i.e. latitude and longitude) to at least postcode-level accuracy and/or a corresponding Australian Business Number (ABN) for around 65% of all applicants in the extracted data set.  Where IPGOD included geolocation coordinates I used these to identify the closest post office and corresponding post code.  Where IPGOD included only an ABN, I used the Australian Business Register (ABR) Web Services API to look up the corresponding post code of the business entity.  I additionally used the Google Maps Geocoding API to try to look up the applicant using its name and any available location information (which is essentially the programmatic equivalent of typing a business name into the Google Maps search box).  Finally, and as an absolute last resort, I attempted to extract a state and postcode for an applicant from the AusPat record of the most recent corresponding patent application.  Through these sources I was able to obtain a postcode for 92% of applicants in my extracted data set, which were then used to generate the map.

It is also necessary to say a few words about the geographic location I have assigned to each postcode.  Each Australian postcode can cover a number of suburbs and/or post offices or distribution centres.  In remote regions, a postcode can cover many tens of thousands of square kilometres!  Some postcodes are used only for post office boxes in particular post offices.  There is, therefore, no unique mapping between a business location, a postcode and a suburb, region or post office.  My solution to this ambiguity was to associate a unique location with each postcode, defined as the geographic average of all Australia Post locations (post offices, distribution centres and/or PO box locations) corresponding with the postcode.  Then, in order to obtain a unique region, suburb or office name to associate with the postcode I identified the Australia Post location that is closest to the geographic mean point.  This will hopefully explain why the names selected in some regions may appear somewhat arbitrary – although, in fact, they are not arbitrary at all, and they make perfect sense to a computer!

The Top 50 Postcodes for Australian Innovation

The list below summarises the data associated with the 50 markers appearing in the map above.  In many cases it will be obvious to readers familiar with the localities why they are so prominent.  Many of them are in or around major academic or research institutions.  Number 2 on the list, the Canberra suburb of Reid, is the location which, by the algorithm noted above, has the benefit of being associated with all applications filed in the name of the Commonwealth Scientific & Industrial Research Organisation (CSIRO), which also featured prominently in my biotechnology map last month.  The reasons for the prominence of some other areas may be less obvious, however I would welcome any observations in the comments below.

#	P'code	Region, Suburb or Post Office	Prov'l	Complete
1	1710	Epping, NSW	48	759
2	2612	Reid, ACT	242	464
3	2113	Blenheim Road, NSW	151	434
4	2000	Sydney, NSW	251	352
5	4000	Brisbane Adelaide Street, QLD	257	334
6	3000	Melbourne, VIC	130	254
7	4067	St Lucia South, QLD	148	168
8	2015	Alexandria, NSW	134	154
9	3122	Kooyong, VIC	59	171
10	2060	McMahons Point, NSW	115	146
11	3095	Eltham, VIC	31	147
12	1221	Royal Exchange, NSW	44	137
13	3121	Richmond, VIC	80	124
14	2006	The University Of Sydney, NSW	128	104
15	3168	Monash University, VIC	64	118
16	2066	Riverview, NSW	103	106
17	4113	Runcorn, QLD	29	123
18	5000	Adelaide, SA	61	109
19	3195	Waterways, VIC	28	116
20	3010	University Of Melbourne, VIC	85	99
21	3130	Laburnum, VIC	22	114
22	3800	Monash University, VIC	127	87
23	8009	Flinders Lane, VIC	38	108
24	4064	Paddington, QLD	90	94
25	4814	Aitkenvale, QLD	105	85
26	3049	Melbourne Airport, VIC	14	104
27	3166	Oakleigh, VIC	30	100
28	6005	West Perth, WA	51	93
29	3178	Rowville, VIC	28	97
30	1455	Botany, NSW	58	87
31	2170	Liverpool, NSW	32	88
32	2867	Long Point, NSW	115	67
33	2127	Homebush Bay, NSW	44	84
34	3175	Bangholme, VIC	33	83
35	3207	Hampton North, VIC	51	78
36	3005	World Trade Centre, VIC	19	85
37	2019	Ermington, NSW	27	83
38	5007	Hindmarsh, SA	14	86
39	2065	St Leonards, NSW	61	74
40	2077	Asquith, NSW	73	69
41	2250	Somersby, NSW	62	70
42	4074	Middle Park, QLD	53	72
43	2145	Wentworthville, NSW	61	68
44	4207	Mount Warren Park, QLD	28	76
45	3068	Clifton Hill, VIC	19	78
46	2164	Smithfield, NSW	24	75
47	3050	Royal Melbourne Hospital, VIC	44	70
48	2067	Chatswood, NSW	28	73
49	3004	Melbourne, VIC	42	69
50	2010	Darlinghurst, NSW	58	65
		Total	3639	6901

Conclusion – The Tyranny of Distance?

It is often said that modern communications and IT systems make geographic distance irrelevant.  Furthermore, the 2016 Australian Innovation System Report (published annually by the Department of Industry, Innovation and Science) once again highlighted Australia’s poor record of business-research sector collaboration:

Australian industry’s collaboration with higher education and research institutions ranked the lowest of 27 countries in the OECD, both for large businesses and for SMEs... (Section 4.2, p 60)

And yet... the map above strongly suggests that innovation in Australia occurs significantly within ‘clusters’ – that ‘birds of an innovative feather flock together’, if you will.  The Innovation System Report also noted that:

Research institutions are a source of expert knowledge that businesses can leverage in order to innovate more effectively, both in outcomes and cost. Collaboration with research organisations such as CSIRO and universities has been found to more than triple the likelihood of businesses reporting annual productivity growth and increases in other performance measures.

Taking all of this together, it is tempting to hypothesise a correlation between innovation, business growth and proximity to centres of academia and research.  And if that is so, then being at a distance from such innovation centres would represent a competitive disadvantage.  Australia is a large country, with a small and dispersed population relative to its size.  It therefore stands to reason that most businesses are not located in close proximity to relevant academic and research institutes.  Could something as mundane as geography be substantially influencing Australia’s poor record of industry-research collaboration?

Tags: Australia, Innovation, Innovation policy, IPGOD, Patent analytics