Geography of research and urban hierarchy

Transcript

1. GEOGRAPHY OF RESEARCH
   AND URBAN HIERARCHY
   

   View Slide

2. FORTHCOMING WORK
   Chapter "Geography of research and urban hierarchy", to be published in Centrality and
   hierarchy of networks and territories, coordinated by Julie Fen-Chong. © ISTE Editions 2022
   Materials and data associated to the github repository https://github.com/Marion-
   Mai/geo_of_research_and_urban_hierarchy, DOI: 10.5281/zenodo.5973814
   The “Geography and demography” field of the Encyclopedia of Science published by the
   scientific publisher ISTE-Wiley – Editor: Denise Pumain
   

   View Slide

3. KEY QUESTIONS
   What is the link between geography of research and urban hierarchy?
   • To what extent is the distribution of research activities dependent on urban hierarchy?
   • Is city size a major determinant of scientific growth?
   • Is there a relationship between the size of cities and the type of scientific activities that
   take place there?
   

   View Slide

4. MAPPING THE GEOGRAPHY OF RESEARCH
   • R&D employment
   • Publication data
   • Citation counts
   Locales with WoS publications between 2000 and 2013 (articles, reviews & letters)
   Crédit: L. Jégou et M. Maisonobe
   

   View Slide

5. SCALING LAWS
   • A method successfully imported from biology (West et al., 1997) in the 2000s
   • Relation between a spatial distribution and the population size of cities
   • Three regimes: sublinear, linear, superlinear
   

   View Slide

6. Finance, O., Swerts, E. (2020). Scaling Laws in Urban Geography. Linkages with Urban Theories, Challenges and Limitations. In Theories and Models of
   Urbanization: Geography, Economics and Computing Sciences, D. Pumain (Ed.). Springer International Publishing, Cham, 67–96.
   « More than a simple concentration index » (Finance & Swerts, 2020)
   

   View Slide

7. From Pumain et al., Cybergeo, 2006
   

   View Slide

8. View Slide

9. RELATION TO CITY SIZE
   Research is an innovative activity
   Innovative activities concentrates in large metropolitan areas because research productivity
   is higher in large metropolitan areas (Bettencourt et al., 2007)
   VS
   The largest cities became larger because these cities were successful in adopting many
   successive innovation (Evolutionary theory, Pumain et al., 2006)
   

   View Slide

10. RELATION TO CITY SIZE
    • Metropolitan advantage according to Bettencourt et al. 2007
    • Innovation stage according to Pumain et al. 2006
    From Pumain et al., Cybergeo, 2006
    

    View Slide

11. ISSUE WITH THE R&D EMPLOYMENT CATEGORY
    • Does not always include employees in Universities and Hospitals
    • In France, these employees are classified in « Education » and « Health » sectors
    • Using publication data can be a way to overcome this limitation
    • Using publication data also enables to test the relation between productivity per capita
    and population size of cities
    

    View Slide

12. From Nomaler, Frenken & Heimericks, Plos ONE, 2014 – US data
    What about productivity?
    

    View Slide

13. View Slide

14. Marion Maisonobe.
    Géographie des activités
    de recherche et
    hiérarchies urbaines.
    Centralités et hiérarchies
    des réseaux et des
    territoires. Forthcoming.
    ISTE Editions.
    Michel Grossetti, Marion
    Maisonobe, Laurent
    Jégou, Béatrice Milard,
    Guillaume Cabanac.
    Spatial organisation of
    French research from the
    scholarly publication
    standpoint (1999-2017):
    Long-standing dynamics
    and policy-induced
    disorder. EPJ Web of
    Conferences, 2020.
    

    View Slide

15. Proximity between the geography of
    universities and the geography of
    Research in France
    • «The distribution of researchers in higher education, because it is globally linked to that of students
    (correlation coefficient of 0.995!), is increasingly modelled on the upper level of the French urban
    structure. » (Brocard, 1991, p.73) ;
    • During the 1990s, «higher education facilities tended to become as widespread as the facilities in
    which compulsory education is completed » so that in the late 2000s, « all major complete higher
    education hubs are less than two hours' drive from other centres in their region or neighbouring
    regions» (Baron, 2010).
    Madeleine Brocard. La science et les régions : géoscopie de la France. In: RECLUS-La Documentation
    Française. 1991. ; Myriam Baron. Les transformations de la carte universitaire depuis les années
    1960 : constats et enjeux. In: Le Mouvement Social, vol. 233, no. 4, 2010. pp. 93-105
    

    View Slide

16. View Slide

17. THE CRITICAL MASS EFFECT IN QUESTION
    • Following the reports of University Alliance (2009 & 2011), we found no evidence of a
    size effect regarding the spatial distribution of scientific activities between urban areas
    (Grossetti et al., 2015, Handbook of geographies of innovation)
    • The spatial distribution of academics explains the spatial distribution of scientific
    activity (volume of publications per urban area) :
    • In France with a very good R2 (95%) - Grossetti et al., 2020
    • In the UK with a very good R2 (88%) - Maisonobe, forthcoming
    → Agglomeration perimeters shared in Maisonobe, Jégou & Eckert, 2018, Cybergeo
    

    View Slide

18. P. A. Balland, C. Jara-
    Figueroa, S. G. Petralia, et
    al. Complex economic
    activities concentrate in
    large cities. In: Nat Hum
    Behav, 4, 2020. pp. 248–
    254
    Does complexity concentrate in metropolitan areas?
    

    View Slide

19. View Slide

20. View Slide

21. View Slide

22. DISCUSSION
    • Influence of urban perimeters (Arcaute et al., 2015)
    • Regression method (OLS) (Leitão et al., 2016) + zero values (Finance & Cottineau, 2019)
    • Residuals (low R2)
    • Other determinants than the city size (spatial logics by institution)
    • Other ways of measuring a spatial concentration and its evolution
    • Promising method: a dominance tree approach to systems of cities (Louail & Barthelemy, 2022)
    

    View Slide

23. SPATIAL LOGICS BY INSTITUTION
    • Universities
    • Town & gown relationships
    • HER decentralization policies
    • Hospitals
    • Research centers
    • Sectoral choices (i.e. space science)
    • Observatories
    • Stations and instruments (fieldwork geography)
    

    View Slide

24. Marion Maisonobe & Bastien Bernela. Exploring the borders of
    a transregional knowledge network. The case of a French
    research federation in green chemistry. International
    Conference on Scientometrics and Informetrics (ISSI 2019), Sep
    2019, Rome, Italy.
    

    View Slide

25. Évolution de la répartition des citations reçues par les publications parues entre 2000 et 2010.
    Measuring spatial concentration dynamics
    

    View Slide

26. CHANGE IN THE GLOBAL CONCENTRATION OF
    PRODUCTION BY CLASSES OF CITIES
    Most publishing cities 2000* 2003* 2007* 2010* 2013* Trend
    Top 10 17.1 15.8 14.7 14.0 14.1
    Top 20 24.6 23.4 22.2 21.3 21.6
    Top 30 30.2 29.0 27.5 26.6 27.1
    Top 50 39.1 37.7 36.0 35.1 35.6
    Top 100 52.8 51.3 49.8 48.7 49.2
    Top 200 69.7 68.3 66.7 65.3 65.1
    Top 500 89.6 88.4 86.7 85.0 84.4
    Top 1000 96.7 96.3 95.5 94.6 94.2
    Total 100 100 100 100 100
    Share of the global total of publications (%)
    Source: Science Citation Index Expanded (articles, reviews and letters)
    *mobile average over three years
    

    View Slide

27. Maisonobe, Marion. «Regional
    Distribution of Research: The
    Spatial Polarization in
    Question». In Handbook
    Bibliometrics, par Rafael Ball,
    377-96. De Gruyter Saur, 2020.
    https://doi.org/10.1515/97831106
    46610-036
    

    View Slide

28. ELABORATING ON OKABE & SADAHIRO’S WORK
    A. Identify the local centers from the
    initial Voronoi tessellation
    B. Draw new Voronoi polygons by
    considering these local centers only
    C. Determine the local centers, etc.
    A. In the end we obtain a dominance
    tree representing the spatial
    hierarchy of the system.
    Each node is characterized
    by its height h in the tree
    Louail & Barthelemy, 2022
    

    View Slide

29. APPLICATION: FRENCH AND US SYSTEMS OF CITIES (1880-2010)
    INSEE population data of French municipalities (since 1876)
    Compilation of US cities populations between 1790 and 2010 (every 10 years) Data come primarily from
    the US Census Bureau
    https://github.com/cestastanford/historical-us-city-populations
    Both datasets are public and available for free
    Louail & Barthelemy, 2022
    

    View Slide

30. POLICY IMPLICATIONS
    • Policy makers firmly believe in agglomeration effects favouring the metropolitan areas:
    « As the share of highly educated people tends to be larger in bigger cities, the productivity
    effects of city size and human capitals can thus reinforce each other » OECD, The Metropolitan
    Century, 2015, https://doi.org/10.1787/9789264228733-en
    • Demonstrating the lack of critical mass effects has important consequences as it
    invalidates the benefit of concentration of research funds and excellence policies targeting
    the biggest cities
    

    View Slide

31. The World is spiky
    Florida, 2005
    « The concentration of creative talents in a few hotspots able to connect to
    the global system of cities is intensifying from the 1990s » (Florida, 2005)
    

    View Slide

32. Maisonobe, M., Jégou, L., & Cabanac, G., Peripheral Forces, Nature Index 563, S18-S19 (2018)
    https://www.nature.com/articles/d41586-018-07210-6
    

    View Slide

33. The GeoScimo website in french and english
    GEOgraphie de la production SCIentifique MOndiale
    URL : https://www.irit.fr/netscity
    An online tool (beta version) to analyse
    and map contemporary scientific
    networks at the city level
    

    View Slide