High Speed Rail methodology hero image


We first explain the approach and general assumptions used in our assessment, followed by details on how we derived the cost and benefit estimate for each component.

General Approach and Assumptions

To estimate the net present value of China’s high-speed rail (HSR) network that is operational at the end of 2019, we used the discounted cash flow approach, the most commonly used valuation method. We first forecast the annual cash flow from the entire HSR network through 2050, then converted it to the 2020 value using a 5% discount rate. All costs and benefits are expressed in 2020 values.

Although HSR is designed for a lifetime operation of 50 years, the cost/benefit analysis is for a 30-year timeframe through 2050. This is because the longer the time horizon, the less certain the estimate, particularly if technologies like magnetic levitation displace current HSR technology in that time period.

Moreover, the benefits of HSR are future oriented. For example, in our forecast, HSR passenger traffic will not peak until 2035. So if the timeframe is set at 10 years, that would likely underestimate the potential realized benefits of HSR.

With that general framework in mind, we made the following assumptions to forecast the annual cash flow through 2050:

HSR Traffic Density: Peaks at 30 million/km by 2035 and levels off after that point.

This is an increase from the current level of 23 million/km and is between the traffic densities of France (25 million/km) and Japan (36 million/km). This 30 million/km assumption is somewhat conservative since it is only 40% of the current passenger density of the Beijing-Shanghai line.

With a third of the routes just completed between 2018-2019, these newer routes tend to see double-digit growth in traffic in the first few years of operation. Combined with the fact that China’s population density is much higher than France’s, the former should easily surpass the latter’s demand for HSR.

Passenger volume: Peaks at 1.06 trillion passengers/km in 2035. This is derived by simply multiplying the traffic density with the total length of the HSR network at the end of 2019.

Shift from other modes of transportation:

  • ~50% of passengers shift from conventional rail
  • 20% of passengers shift from air travel
  • 15% of passengers shift from bus and car

We use the World Bank’s (WB) 2015 estimates for each mode of transportation and assume that these substitution effects last through 2025. Beyond 2025, all HSR growth represents new traffic rather than substitution effects.

The magnitude of shifts from different modes of transport is crucial for estimating HSR’s cost and benefit. For instance, when HSR substitutes for air travel, that leads to cost savings because the operating cost of the latter is more than twice that of the former. In contrast, shifting from conventional rail will be costlier because the operating cost of HSR is slightly higher.

Annual discount rate: 5% for both costs and benefits. We assume this discount rate because the borrowing cost for HSR investment is currently less than 4% and has been below 5% for most of the last decade.

Calculating costs and benefits

Below we detail more specifically how we arrived at the costs and benefits of HSR over the 30-year timeframe. These are all generally accepted factors of cost and benefit for HSR, and where appropriate, we used the WB’s estimate for our calculations.

One estimate where we differed from the WB is that of HSR’s ROI: 6.5% (our estimate) vs. 8% (WB’s estimate). This is the result of having more conservative assumptions in our estimate. First, we did not include positive externalities such as agglomeration effects. Second, we have a lower residual value of HSR, which is equivalent to assuming a higher depreciation rate. Third, we assume that HSR will never fully repay debt and will continue making interest payments for the entire 30-year period.


Construction Cost: Based on WB’s estimate of 130 million yuan/km per year multiplied by the estimated length of new HSR under construction each year.

Total construction cost is the sum of annual construction cost after subtracting the residual value of the HSR network in 2050. We subtract the residual value from construction cost because the HSR network will still be in use after 2050 and has a positive value. Put differently, what is included can be interpreted as the total depreciation of the network through 2050. We assume the residual value to be 30% of the HSR construction cost, a more conservative assumption compared to what is likely closer to 40% in reality.

Interest Payment: 4% on HSR-related debt (all HSR investments are assumed to be financed by equal portions of debt and equity).

Maintenance Cost: Based on WB’s higher-end estimate of HSR maintenance cost of 2.3 million yuan/km per year.

Rolling Stock: Based on annual spending on rolling stock, or the actual trains, before 2019 and based on an imputed annual usage cost after 2019. Usage cost is calculated by multiplying annual traffic with usage cost per train car (each train car is assumed to have an occupancy of 60 people). The per km train car usage cost is 7.5 yuan.

Operating Cost: Based on the estimated operating cost of the Beijing-Shanghai line, the only route with detailed operating cost data, minus the rolling stock cost.


Air Travel Substitution: Estimated based on expected annual savings from the 20% passenger traffic shift to HSR. On a per passenger-km basis, travel by HSR is about 50% cheaper than by plane, which implies that transporting the same number of passengers by HSR requires only half the upfront capital investment relative to air travel.

Lower Investment for Conventional Rail: With the shift to HSR, China can accommodate growing travel demand without investing as much in conventional rail. Since conventional rail costs less than HSR to build, deferred conventional rail construction is assumed to be a quarter of the HSR construction cost.

Time Savings: Estimated based on the 20% passenger traffic shift from airplane to HSR, multiplied by the average hourly wage of an urban worker.

For distances no greater than 800 km, or roughly three hours of travel by HSR, it saves time to take HSR over an airplane. For instance, the “first mile” and “last mile” of air travel take much longer than HSR because of security checks at the airport and the fact that most airports are farther away from final destinations in urban centers.

Operating Cost Savings: Estimated based on the average operating costs of various modes of transportation and the passenger traffic shift assumptions specified above (e.g. HSR’s operating cost is higher than conventional rail but much lower than airplanes).

Generated Traffic Benefits: Estimated based on time spent traveling on HSR, multiplied by the average hourly wage of an urban worker.

In theory, the value of the trip itself must be greater than the value of time spent traveling, or else one would not bother wasting their valuable time. But to stick with our more conservative assumptions, in this case we assume the generated traffic benefit is simply equal to the value of time spent traveling. If agglomeration effects are included, the benefit would be much greater. But given the lack of consensus on the magnitude of agglomeration, we did not account for these positive externalities in our estimate.


Project leadsDamien Ma, Houze Song
DevelopmentChris Roche
Creative leadsAnnie Cantara, Yna Mataya
Research supportYa-han Cheng