というNBER論文が上がっている（ungated版）。原題は「A Comprehensive GIS Database for China's Surface Transport Network with Implications for Transport and Socioeconomics Research」で、著者はSteven J. Davis（スタンフォード大）、Meijun Qian（中国人民銀行）、Wen Zeng（オーストラリア国立大）。
以下はその要旨。
We build a granular Geographic Information System (GIS) database that covers China’s national highways, modern motorways, traditional railways, high-speed railways, and waterways at an annual frequency from 1993 to 2020. After describing the database and its construction, we characterize the development of China’s surface transport network. Overall network length more than tripled after 1993. Modern motorways and high-speed railways, nearly non-existent in the early 1990s, account for half the increase in network length. The average distance between county centroids and transport access points fell by more than half, from 13.28 km in 1993 to 5.81 km in 2020. Common county-level measures of connectivity and centrality rose by 22% and 115%, respectively. We also show that discrepancies between distance to motorway access and straight-line distance to motorway routes are often large, and the discrepancies correlate with calendar time, terrain features, and economic development. Because motorway access is vital to local economic development in China, this finding raises the need to re-assess previous research that uses straight-line distance to proxy actual distance. Our GIS database is freely available on an open-access basis, creating an empirical laboratory for new research in multiple directions.
（拙訳）
我々は、中国の国道、現代的な高速道路、従来型の鉄道、高速鉄道、および水路をカバーした1993年から2020年までの年次の詳細な地理情報システム（GIS）データベースを構築した。データベースとその構造を説明した後に我々は、中国における地表・水面の輸送網の発展の特性を説明する。全輸送網の長さは1993年以降に3倍以上に伸びた。現代的な高速道路と高速鉄道は、1990年代初期にはほぼ存在しなかったが、輸送網の伸びの半分を占めた。県の中心と輸送アクセスポイントとの平均的な距離は1993年の13.28kmから2020年の5.81kmへと半分以下となった。標準的な県レベルの接続性と中心性の指標はそれぞれ22%、115%増加した。我々はまた、高速道路の入口への距離と高速道路ルートへの直線距離との乖離が大きいことが多く、その乖離は暦年、地域の特性、および経済発展と相関していることを示す。中国では地域経済の発展にとって高速道路へのアクセスは極めて重要であるため、この発見は実際の距離の代理変数として直線距離を用いたこれまでの研究の再評価を迫るものである。我々のGISデータベースはオープンアクセスとして無料で利用可能であり、様々な方向への新たな研究について実証的な研究室を提供するものである。
以下は高速道路の発展を示した論文の図。
以下は指標の定義を示した論文の表。
| Measure | Definition |
| Panel A: Location-specific measures | |
|---|---|
| Access indicator | Whether a location has access to transport within a distance |
| Access distance | The shortest distance to the nearest access point, i.e., motorway entrance/exit, railway station, or highway route |
| Centrality (Betweenness index) | Ratio of the number of links summed over all shortest paths in the network that pass through a county to the total number of links on the shortest paths summed over all node pairs in the network |
| Centrality (Degree index) | Ratio of the number of nodes connecting a county to the total number of nodes in the network |
| Centrality (Closeness) | Ratio of the total number of nodes in a network to the number of links along the shortest paths between a county and all other nodes |
| Panel B: Pairwise measures | |
| Travel Distance | Travel distance between locations via transport mode from one city to another or from a postal code centroid to a port |
| Travel Time | Shortest travel time between geolocations with a transport mode |
| Panel C: Network-level measures | |
| Centrality | Aggregations of location-specific centrality measures in a region |
| Route density | Transport length divided by its area in a region |
| Node density | Number of access points divided by its area in a region |
| Connectivity (Beta index) | Average number of links per node in a network (total number of links divided by total number of nodes |
| Connectivity (Gamma index) | Number of links in a network divided by the maximum feasible number |
Note: This table lists all location-specific, pairwise, and network-level transport measures, as well as their definitions.
