Under de senaste decennierna, stadsbefolkningen i Kinas städer har vuxit avsevärt, och stigande inkomster har lett till en snabb expansion av bilägandet. Verkligen, Kina är nu världens största marknad för bilar. Kombinationen av urbanisering och motorisering har lett till ett akut behov av transportpolitik för att ta itu med urbana problem som trängsel, luftförorening, och utsläpp av växthusgaser.

De senaste tre åren, ett MIT -team som leds av Joanna Moody, forskningsprogramledare för MIT Energy Initiatives Mobility Systems Center, och Jinhua Zhao, Edward H. och Joyce Linde docent vid Institutionen för stadsstudier och planering (DUSP) och chef för MIT:s JTL Urban Mobility Lab, har undersökt transportpolitik och beslutsfattande i Kina. "Det antas ofta att transportpolitiken i Kina dikteras av den nationella regeringen, "säger Zhao." Men vi har sett att den nationella regeringen sätter upp mål och sedan tillåter enskilda städer att bestämma vilken politik de ska genomföra för att nå dessa mål. "

Många studier har undersökt transportpolitik i Kinas megastäder som Peking och Shanghai, men få har fokuserat på de hundratals små och medelstora städer som finns över hela landet. Så lynnigt, Zhao, och deras team ville överväga processen i dessa förbisedda städer. Särskilt, de frågade:hur bestämmer kommunala ledare vilken transportpolitik som ska genomföras, och kan de bättre kunna lära av varandras erfarenheter? Svaren på dessa frågor kan ge vägledning till kommunala beslutsfattare som försöker ta itu med de olika transportrelaterade utmaningarna som deras städer står inför.

Svaren skulle också kunna bidra till att fylla en lucka i forskningslitteraturen. Antalet och mångfalden av städer över hela Kina har gjort det svårt att utföra en systematisk studie av stadstransportpolitik, Ändå blir det ämnet allt viktigare. Som svar på lokal luftföroreningar och trafikstockningar, vissa kinesiska städer antar nu policyer för att begränsa bilägande och användning, och dessa lokala policyer kan i slutändan avgöra om den oöverträffade tillväxten i rikstäckande försäljning av privata fordon kommer att bestå under de kommande decennierna.

Policyinlärning

Transportpolitiker världen över drar nytta av en praxis som kallas policy-learning:Beslutsfattare i en stad ser till andra städer för att se vilken politik som har och inte har varit effektiv. I Kina, Peking och Shanghai ses vanligtvis som trendsättare inom innovativ transportpolitik, och kommunledare i andra kinesiska städer vänder sig till dessa megastäder som förebilder.

Men är det ett effektivt tillvägagångssätt för dem? Trots allt, deras urbana miljöer och transportutmaningar är nästan säkert helt olika. Vore det inte bättre om de såg till att "jämföra" städer som de har mer gemensamt med?

Lynnig, Zhao, och deras DUSP-kollegor – postdoc Shenhao Wang och doktorander Jungwoo Chun och Xuenan Ni, alla i JTL Urban Mobility Lab—hypoteserade ett alternativt ramverk för policyinlärning där städer som delar en gemensam urbaniserings- och motoriseringshistoria skulle dela med sig av sin policykunskap. Liknande utveckling av stadsrum och resmönster skulle kunna leda till samma transportutmaningar, och därför till liknande behov av transportpolitik.

För att testa deras hypotes, forskarna behövde ta upp två frågor. Att börja, de behövde veta om kinesiska städer har ett begränsat antal gemensamma urbaniserings- och motoriseringshistorier. Om de grupperade de 287 städerna i Kina baserat på dessa historier, skulle de sluta med ett måttligt litet antal meningsfulla grupper av jämställda städer? Och för det andra, skulle städerna i varje grupp ha liknande transportpolicyer och prioriteringar?

Gruppera städerna

Städer i Kina är ofta grupperade i tre "nivåer" baserat på politisk administration, eller de typer av jurisdiktionsroller som städerna spelar. Nivå 1 inkluderar Peking, Shanghai, och två andra städer som har samma politiska befogenheter som provinser. Tier 2 includes about 20 provincial capitals. The remaining cities—some 260 of them—all fall into Tier 3. These groupings are not necessarily relevant to the cities' local urban and transportation conditions.

Moody, Zhao, and their colleagues instead wanted to sort the 287 cities based on their urbanization and motorization histories. Fortunately, they had relatively easy access to the data they needed. Varje år, the Chinese government requires each city to report well-defined statistics on a variety of measures and to make them public.

Among those measures, the researchers chose four indicators of urbanization—gross domestic product per capita, total urban population, urban population density, and road area per capita—and four indicators of motorization—the number of automobiles, taxis, buses, and subway lines per capita. They compiled those data from 2001 to 2014 for each of the 287 cities.

The next step was to sort the cities into groups based on those historical datasets—a task they accomplished using a clustering algorithm. For the algorithm to work well, they needed to select parameters that would summarize trends in the time series data for each indicator in each city. They found that they could summarize the 14-year change in each indicator using the mean value and two additional variables:the slope of change over time and the rate at which the slope changes (the acceleration).

Based on those data, the clustering algorithm examined different possible numbers of groupings, and four gave the best outcome in terms of the cities' urbanization and motorization histories. "With four groups, the cities were most similar within each cluster and most different across the clusters, " says Moody. "Adding more groups gave no additional benefit."

The four groups of similar cities are as follows:

• Cluster 1:23 large, tät, wealthy megacities that have urban rail systems and high overall mobility levels over all modes, including buses, taxis, and private cars. This cluster encompasses most of the government's Tier 1 and Tier 2 cities, while the Tier 3 cities are distributed among Clusters 2, 3, and 4.
• Cluster 2:41 wealthy cities that don't have urban rail and therefore are more sprawling, have lower population density, and have auto-oriented travel patterns.
• Cluster 3:134 medium-wealth cities that have a low-density urban form and moderate mobility fairly spread across different modes, with limited but emerging car use.
• Cluster 4:89 low-income cities that have generally lower levels of mobility, with some public transit buses but not many roads. Because people usually walk, these cities are concentrated in terms of density and development.

City clusters and policy priorities

The researchers' next task was to determine whether the cities within a given cluster have transportation policy priorities that are similar to each other—and also different from those of cities in the other clusters. With no quantitative data to analyze, the researchers needed to look for such patterns using a different approach.

Först, they selected 44 cities at random (with the stipulation that at least 10 percent of the cities in each cluster had to be represented). They then downloaded the 2017 mayoral report from each of the 44 cities.

Those reports highlight the main policy initiatives and directions of the city in the past year, so they include all types of policymaking. To identify the transportation-oriented sections of the reports, the researchers performed keyword searches on terms such as transportation, road, bil, bus, and public transit. They extracted any sections highlighting transportation initiatives and manually labeled each of the text segments with one of 21 policy types. They then created a spreadsheet organizing the cities into the four clusters. Till sist, they examined the outcome to see whether there were clear patterns within and across clusters in terms of the types of policies they prioritize.

"We found strikingly clear patterns in the types of transportation policies adopted within city clusters and clear differences across clusters, " says Moody. "That reinforced our hypothesis that different motorization and urbanization trajectories would be reflected in very different policy priorities."

Here are some highlights of the policy priorities within the clusters:

The cities in Cluster 1 have urban rail systems and are starting to consider policies around them. Till exempel, how can they better connect their rail systems with other transportation modes—for instance, by taking steps to integrate them with buses or with walking infrastructure? How can they plan their land use and urban development to be more transit-oriented, such as by providing mixed-use development around the existing rail network?

Cluster 2 cities are building urban rail systems, but they're generally not yet thinking about other policies that can come with rail development. They could learn from Cluster 1 cities about other factors to take into account at the outset. Till exempel, they could develop their urban rail with issues of multi-modality and of transit-oriented development in mind.

In Cluster 3 cities, policies tend to emphasize electrifying buses and providing improved and expanded bus service. In these cities with no rail networks, the focus is on making buses work better.

Cluster 4 cities are still focused on road development, even within their urban areas. Policy priorities often emphasize connecting the urban core to rural areas and to adjacent cities—steps that will give their populations access to the region as a whole, expanding the opportunities available to them.

Benefits of a "mixed method" approach

Results of the researchers' analysis thus support their initial hypothesis. "Different urbanization and motorization trends that we captured in the clustering analysis are reflective of very different transportation priorities, " says Moody. "That match means we can use this approach for further policymaking analysis."

At the outset, she viewed their study as a "proof of concept" for performing transportation policy studies using a mixed-method approach. Mixed-method research involves a blending of quantitative and qualitative approaches. In their case, the former was the mathematical analysis of time series data, and the latter was the in-depth review of city government reports to identify transportation policy priorities. "Mixed-method research is a growing area of interest, and it's a powerful and valuable tool, " says Moody.

She did, dock, find the experience of combining the quantitative and qualitative work challenging. "There weren't many examples of people doing something similar, and that meant that we had to make sure that our quantitative work was defensible, that our qualitative work was defensible, and that the combination of them was defensible and meaningful, " hon säger.

The results of their work confirm that their novel analytical framework could be used in other large, rapidly developing countries with heterogeneous urban areas. "It's probable that if you were to do this type of analysis for cities in, säga, Indien, you might get a different number of city types, and those city types could be very different from what we got in China, " says Moody. Regardless of the setting, the capabilities provided by this kind of mixed method framework should prove increasingly important as more and more cities around the world begin innovating and learning from one another how to shape sustainable urban transportation systems.

Den här historien återpubliceras med tillstånd av MIT News (web.mit.edu/newsoffice/), en populär webbplats som täcker nyheter om MIT-forskning, innovation och undervisning.