Measuring road and rail infrastructure quality in China

By Jovina Ang

SMU Office of Research – China has been investing heavily in infrastructure development in the past 40 years.

The amount spent on infrastructure development has ranged from 4.8 percent of GDP to as high as 24 percent of GDP, which was way higher than the combined amount spent in North America and Western Europe.

Infrastructure development was pivotal for the Chinese government to generate growth and modernise the economy.

The strategy was so successful that the country prospered with an average GDP growth of nine percent per year and more than half its population, or 770 million people, were lifted out of poverty.

“While the GDP figures show the importance of infrastructure development for driving economic growth, it is unclear how it has affected travel time, as well as other economic factors such as wages in the different locations in China,” observes Ma Lin, SMU Assistant Professor of Economics, Lee Kong Chian Fellow and Fellow of the SMU Urban Institute.

The construction of roads and railroads is not without challenges. 

A third of China is hilly and mountainous. Its western region bordering India, Nepal and Central Asia is more than 4,000 meters above sea level. Compare this to its eastern terrain adjacent to the Pacific Ocean where the lay is approximately 200 meters above sea level.

“Due to the geographical terrain, lower design speeds are incorporated in the design of roads and railroads in the hilly and mountainous areas. It is also likely that the infrastructure in these areas is built of a lower quality with lower load bearing and lower design speeds,” Professor Ma continued.

“Previous research on infrastructure development has focused primarily on showing point-to-point binary connectivity or measuring connectivity at the pixel level,” he added, referring to research establishing the existence of links without taking into consideration factors such as terrain or travel speed.

While pixel displays provide a visual representation of infrastructure development over time, it is not comprehensive enough to ascertain the design speeds across the infrastructure networks.

Another inherent flaw of using pixels to measure infrastructure development lies in the assumption that all roads and railroads are built to the same quality, and therefore, provides the same level of speed connectivity regardless of terrain or when it was built.

With many innovations and advancements in road and rail technology, recently constructed roads and railroads have been shown to yield faster speeds and have better load-bearing compared to those networks that were constructed 30 to 40 years ago.

The quality of connectivity matters as economists have shown that higher quality roads are directly correlated with higher household consumption and income levels.

Having grown up with construction sites all around him, Professor Ma also wanted to learn more about the quality of infrastructure development in China. 

He said: “When I was growing up in Beijing, there was not a day when I did not see a construction site. I have also been amazed by how fast and how good the infrastructure networks have become in the past 20 years.”

The research

Despite the rapid development of infrastructure across China, there exists no systematic documentation that details the quality of it using design speeds.

Because of this gap, Professor Ma, together with his co-researcher, Associate Professor Tang Yang from Nanyang Technological University, decided to embark on the research project.

The Research Methodology

In order to build a comprehensive dataset at the pixel level that shows the design speeds of roads and railroads for a specific year, the research team developed the following steps:

• Step 1: Obtain and digitise all the road and railroad infrastructure networks from the official publications from the Ministry of Transportation in China including Transportation Yearbooks, provincial map collections, and the Chronicles of Railroad Construction, dating as far back as 1881 when the first railroad was constructed by the Qing Dynasty.
• Step 2: Detailing the variations in infrastructure quality using the standard engineering design codes to infer the design speeds. Different design codes were used to depict the different design speeds of the infrastructure, e.g., higher-rated speed roads versus lower-rated speed roads, rugged terrains versus plains.
• Step 3: Break down the vast transportation networks into the different segments so that each segment could be categorised by the year of construction with its corresponding design speeds.
• Step 4: Determining the terrain type for each segment at the pixel level so that the appropriate design speeds could be assigned.

In so doing, the research team could ascertain the design speeds for the infrastructure networks and measure the distance between two pixels by design speeds.

The results

The results showed a significant reduction in travel time for infrastructure built from 1994 to 2017.

On the whole, the travel time on roads reduced by 32 percent. However, when the design speeds were taken into consideration, the median reduced to 10 percent because the newly constructed roads and highways are typically located in the mountainous regions with low design speeds.

As for rail transportation, the travel time reduced by 37 percent for freight travel and 59 percent for passenger travel. When the design speeds were omitted in the calculation, the median travel time reduced by 46 percent for freight travel and 61 percent for passenger travel.

The results also show that the aggregate welfare of China’s population improved by 57 percent as there was reduced internal trade frictions and better allocation of people across the distances.

Contribution to research

Even though this model of dataset was developed to depict the quality of infrastructure in China, the same methodology can be applied to ascertain the quality of road and rail infrastructure in other parts of the world.

Professor Ma’s paper was published in the Journal of International Economics and can be found at this link.

Back to Research@SMU November 2024 Issue