Page 378 - 360.revista de Alta Velocidad - Nº 6

Yang, Haoran. Dobruszkes, Frédéric. Wang, Jiao’e. Dijst, Martin.

class in the HSR networks to the fifth class in the airline networks, while Beijing-Shenzhen and
ShanghaiShenzhen upgraded from the fifth class in HSR networks to the second class in airline
networks. Therefore, it can be found that although for the city links with both HSR and airline
connections, end nodes of links are mainly the major cities or hubs, whether they are dominant
on HSR or airline networks are largely based on their geographical distances between the city
ends.
4.1.3 City and link strength vs. attributes of urban systems

To interpret aforementioned results, we performed a multiple linear regression to investigate
potential factors of both city and link strength

Table 2. Multiple regression on city strength

a
a
DIT_HSR DIT_Airline
Standardized coefficients Standardized coefficients

GDP per capita 0.576*** 0.184***
a
Average distance 0.022 0.299***
a
Population 0.414*** 0.141***
a
Administrative level 0.185** 0.502***

Observations 105 168

Adjusted R-squared 0.689 0.665

* p<0.1 ** p<0.05 *** p<0.01
Ln transformation.
a

Table 2 shows results at the city level, it appears that GDP per capita and population of cities are
the first and second most significant indicators for the city strength in HSR networks, compared
to the administrative level of cities and the average distance to others in airline networks.
Cities in HSR networks show higher elasticities of GDP per capita and population to the city
strength than in airline networks, meaning that compared to airline travel, HSR travel is still
mainly concerned about the connections to cities with higher socio-economic performance.
The positive sign of the average distance to other cities in airline networks rather than in HSR
networks indicates if one city is far away from others, the airline travel becomes a more suitable
transportation alternative than the HSR travel for middle and long haul travel. Furthermore,
the negative coefficients of the administrative level in both transportation networks indicate
that in general the higher the city’s position in the administrative hierarchy, the more likely
passengers need to travel either to/from other cities. However, city strength is much sensitive
to the administrative level in the airline network than in the HSR network. Therefore, it could
expect both distance and administrative hierarchy more affect the city strength in airline than
HSR networks.

376 360.revista de alta velocidad