U.S. Energy Information Administration logo
Skip to sub-navigation

Analysis & Projections

Issues in International Energy Consumption Analysis: Chinese Transportation Fuel Demand

Release date: July 1, 2014


Since the 1990s, China has experienced tremendous growth in its transportation sector. By the end of 2010, China's road infrastructure had emerged as the second-largest transportation system in the world after the United States. Passenger vehicle sales are dramatically increasing from a little more than half a million in 2000, to 3.7 million in 2005, to 13.8 million in 2010.1 This represents a twenty-fold increase from 2000 to 2010. The unprecedented motorization development in China led to a significant increase in oil demand, which requires China to import progressively more petroleum from other countries, with its share of petroleum imports exceeding 50% of total petroleum demand since 2009.2  In response to growing oil import dependency, the Chinese government is adopting a broad range of policies, including promotion of fuel-efficient vehicles, fuel conservation, increasing investments in oil resources around the world, and many others.

In recent years, oil demand forecasting has been gaining prominence as a tool utilized by the Chinese government in preparation of Five-Year Plans and long range planning. Several government agencies in China are tasked with providing oil demand forecasts to aid in formulating long-term policies. The Energy Research Institute (ERI) of the National Development and Reform Commission (NDRC) carries out energy supply and demand forecasting. The Chinese Academy for Social Sciences (CASS) is a think tank affiliated with the State Council, who conducts economic research on oil in China. China's largest national oil company—China National Petroleum Corporation (CNPC)—forecasts oil supply and demand via the CNPC Research Institute of Economics and Technology. Although these forecasts are not directly from the government, they are quite influential in China.

Since the transportation sector now accounts for the largest share of petroleum demand and is projected to remain the driving force of oil consumption growth in the long term (having recently surpassed industrial petroleum demand), transportation demand modeling has been given increased attention in China in recent years. The regulatory provisions of the Law of the People's Republic of China on Urban and Rural Planning require all cities to regularly submit city-level master transportation plans, which has prompted many Chinese cities to develop their respective transportation models. Regularly updated comprehensive transportation plans are now a norm in many Chinese cities, with most of these plans based on travel demand models with a land-use component. In addition, many think tanks and research institutions are working on developing national travel demand models.

The most prominent travel demand modeling efforts in China are conducted by several public and semi-public entities including the China Academy of Urban Planning and Design (CAUPD), Transportation Planning Institutes, and Transportation Research Centers in major cities. These entities are the main authorities on travel demand forecasting and are recognized for their strong modeling capabilities. These capabilities stem in part from the entities' access to statistical datasets, which are obtained through relationships with the government. Access to statistical datasets owned or collected by the government is the most significant barrier for other entities' transportation modeling efforts. The primary reasons for government restrictions on data access include concerns over the political implications of data and lack of regulatory requirements for public access, as well as data reliability and accuracy.

Chinese travel demand models are based in large part on western travel demand methodologies, which have a long history (80+ years) and a large volume of accumulated literature, modeling methods, and research. The most commonly used models in China are based on urban planning methodologies and include the 4-step and activity-based models, which utilize software packages such as TransCAD, PTV Group suites, EMME, Citilab's suites, and others. The key variables in these models include mode choice, vehicle ownership, land use, commute distance, geographic files for road/transit networks, employment by zone, chosen travel path, residential and work locations, and socio-economic variables (income, gender, age, etc.). These models are designed to simulate existing travel conditions and forecast future year travel for the entire transportation system of a given city for the transit, auto, and walk/bike modes. Data inputs in these models include household travel surveys, Master Land Use Plans of base and future years, public rider surveys, socio-economic and road traffic surveys, employment surveys, transportation network information, and others.

Travel demand forecasting in China is done primarily for urban and long-range planning to project crude oil and products demand by the transportation sector. Each of these forecasting methods and corresponding models has its own set of drivers, inputs, and variables. Very little information is available on methodology used for long-range planning particularly with respect to Five-Year Plans. Urban planning models are local in nature and cannot be uniformly applied to every city or area to generate a national oil demand forecast based on aggregating cities’ models. There are no known national transportation models as of now, however, several entities are known to be working on their development. Thus, there is both much and little information available about Chinese transportation energy demand: a rich, but difficult-to-access base of urban-level data and forecasting, coupled with scant national-level information. This is a challenging situation when considering the importance of Chinese transportation energy demand for the global outlook.

The key drivers of travel demand considered in the travel demand and transportation models are varied depending on the purpose of the forecast. Urban planning models utilize various surveys, while other forecasting models focus on estimates of long-term vehicle stock, ownership, and saturation levels to generate oil demand view. Yet other models consider travel behavior to be most indicative of the future direction of travel demand and include various behavioral survey data. It would be useful to review the existing transportation methodologies, with a particular focus on methodologies most frequently used by the Chinese forecasters to examine the determinants of travel demand. This could then be related to the need for an internationally consistent methodology for application to China as well as other regions.

Travel demand forecast methodologies

Travel demand forecast methodologies trace their beginnings to the 1950s, when the first models were introduced. These models play a central role in urban and regional transportation modeling. In the last few decades, methods for travel demand forecasting have been expanded and advanced, with two primary modeling approaches gaining most prominence: activity-based and trip-based.

Categorization of multimodal inter-regional travel demand analysis methods

The countries of the European Union utilize advanced large-scale travel models. As the need for analyzing travel between member countries and worldwide becomes increasingly important, several pan-European travel demand models have also been developed, including the MYSTIC project, the STREAM model, and the most recent TRANS-TOOLS project. In addition, aggregate direct demand studies have been conducted to estimate multimodal travel demand patterns in countries outside of the European Union. Integrated national models have also been applied in analyzing travel demand in many European countries.

One recently completed international study has reviewed approximately 60 multimodal inter-regional travel demand models across the world and categorized them into four distinct groups (Figure 1).3 All of these methodologies are capable of estimating multimodal origin-destination matrices (OD), and have produced operational models. The key difference between these methodologies lies in consideration of the travel behavior in response to changing policies.

See full report


1Economist Intelligence Unit (www.eui.com), annual data series "Passenger Vehicle Registrations," and China Association of Automobile Manufacturers (www.caam.org.cn/english/).

2CERA, "China oil data center," annual data series 2002-2013; PFC Energy, "Oil Balance Summary," annual data series, 2008-2014. Proprietory data sets.

3Lei Zhang and Chenfeng Xiong, "Multi-Modal Inter-Regional Travel Demand Estimation: A Review of National-Level Passenger Travel Analysis Methodologies and Their Implications on the Development of National and Statewide Travel Demand Models in the U.S.," Transportation Research Board 2011.