Search interesting materials

Friday, December 14, 2018

Concerns with Delhi Metro

by Shubho Roy and Ajay Shah.

The primary input that goes into an infrastructure project is money. In the case of the Delhi Metro, cost estimates run to Rs.5.52 Billion per km for the underground stretches and Rs.2 Billion per km for the over-ground stretches. The construction cost does not include the cost of land, which the government provides at a subsidised rate.

Turning to the revenues, the arithmetic is clear. The revenue of Delhi Metro comes from the price per ride and the number of rides. To fix intuition, suppose we spend Rs.100 on an asset. Suppose the capital requires annual payments of Rs.10 per year. The infrastructure asset has to then generate at least Rs.10 of cash per year, after paying for the (small) running costs. This cash can come about in many ways, e.g. sell 1 ticket at Rs.10 or sell 10 tickets at Rs.1, and so on.

The cost of capital in India is relatively high, given the presence of high macro/financial risk and a closed capital account. The annual cost of servicing the debt and equity capital, that goes into such projects, will be higher than is the case elsewhere in the world. The costs of building an urban metro system in India are similar to those seen elsewhere in the world. But the same infrastructure asset has to generate a higher revenue stream in India given a higher cost of capital. This means that the user charge or the intensity of use, or both, have to be higher in India.

Hence, to think about the viability of the Delhi Metro, we have to examine the extent to which it beats global benchmarks in user charges or intensity of use or both.

User Charges

User Charges for Metro

As the graph above shows, Delhi Metro has one of the lowest user charges in the world, when compared with other large metro systems. So we're not getting through on this one.

Intensity of use

We could just look at the number of rides in the Delhi Metro. Matters become a little more complicated for international comparisons, as the number of rides varies (slightly) with the length of the metro system. Bigger metro systems get somewhat more rides. To compare apples with apples, we look at how Delhi Metro should have fared in the context of these relationships, as observed worldwide.

Rides per Km

As the graph above shows, on this metric also, the Delhi Metro fares poorly. Delhi Metro is out of line when compared with the relationship seen worldwide. In this graph, the size of the bubble is proportional to total annual rides. Mumbai's ancient train system is faring well in translating kilometres of track into rides. The old Kolkata Metro also does a pretty good job in generating rides per km of track. But Delhi is not.

Thus there is a financial problem

The cost of a metro system in many big cities of the world is similar. The dominant element of this cost is the cost of capital. This is a more significant barrier in India, as the cost of capital is higher in India when compared with most large countries. To make ends meet, a metro system has to achieve the required cash flow by having an adequate user charge multiplied by the number of rides. In Delhi Metro, we are faring poorly on both counts: We don't have an adequate user charge, and we don't have the required number of rides.

Rides and User Charges

As the graph above shows, Delhi fares poorly on both Rides per Km per Year and the average price charged for a commute. The size of the bubble represents the size of the metro system.

Suppose we treat Shanghai as a benchmark. Their charge is similar to that in Delhi, but they get 5.49 million rides per km of track against Delhi's 3.41. There is a gap of about 60%.

How could this gap be closed by holding prices intact? Last year, Delhi's metro system was 296 km long. Delhi would need to get 615.68 million more rides per year without increasing track length, to catch up with Shanghai. That is a 60% rise from Delhi's 2017 ridership of 1007.9 million rides.

Conversely, to obtain the same revenue per km of track, Delhi needs to charge 60% more per ticket. This seems far more feasible.

As Delhi adds more lines next year, the denominator will grow. This will translate into a bigger challenge in terms of obtaining the requisite number of rides.

Positive externalities

We could argue that Delhi Metro is all about positive externalities and that we should not be so worried about the lack of revenues. We can then use other means like taxes to pay for the system. In this vision, the essence of urban infrastructure lies in promoting dense interactions between residents. This is measured by counting the rides per capita. A metro system is inducing large positive externalities if it achieves high values of rides per capita per year.

Rides per Capita riderpercapita

As the evidence above shows, Delhi Metro fares poorly on the rides per capita per year. Delhi is not a small metro system. It now ranks on amongst the top ten metro systems. As the graph shows, in such large systems, the rides per capita should be higher, as the metro network covers the city better.

For an example, Delhi and Madrid are both at about 200 kilometres of line. But in Delhi, there are 25 rides per person per year, while Madrid is at about 100. We are about 4 times worse than we ought to be.


To justify infrastructure assets such as Delhi Metro, we in India have to find exceptionally large user charges, given the difficulties of macro/finance policy which impose an elevated cost of capital. Or, to the extent that the direct revenues are inadequate, we have to justify the expenditure based on an externalities argument.

Delhi Metro is faring poorly on all three fronts. The cost per ride is low, i.e. the user charge is inadequate. The number of rides per year is low. The intensity of usage is low. We have a problem.

We conjecture that part of the problem may lie not in Delhi Metro but in Delhi's urban planning. Delhi is a vertically challenged city, with severe restrictions on the number of floors a person can build. This FSI restriction creates large areas of low-density housing. For example, the IIT Delhi Metro Station is bound by IIT Delhi on one side and the Green Belt on the other. IIT Delhi has a student population of less than 8,000 and the Green Belt is entirely uninhabited for a few kilometres. In effect, the IIT Metro Station does not have a catchment area of an adequate number of potential customers.

The authors would like to thank Vimal Balasubramaniam, Devendra Damle, Ashim Kapoor, and Shalini Mittal for valuable contributions.


  1. does it means that if delhi has more high storeyed building, rides per capita will increase? This is also shown from graph 2 which shows singapore (high storeyed building city).

    However, an another high storeyed city like dubai don't have high per capita metro ride. The primary reasons are culturl (poor people use metro) and availibility of low interest auto loans.

    The dubai reasons could be more true for delhi which is equally class conscous in nature.

    An another reason for high metro ride in singapore is very high cost of car license.

    Hence, in my view, the solution could be predominantly culturual (i.e. an non elitist society) and increasing the cost of car license.

    1. It is not clear that it is cars which cause the problem. Delhi has seen a growth in cars, but it is not out of line. In fact, there were only 118,424 private passenger vehicles in Delhi (6,648,730 bikes), in 2017. Even considering adjoining areas like Gurugram and Noida, that is not an exceptionally large number of cars. Singapore has around 600,000 passenger vehicles.

  2. So, when people ask me about how to think about policy in India the right way, especially young folks, I point them to articles such as these:

    Activism and wonkery are the yin and yang

    Become a public policy thinker in three easy steps

    Hierarchy of thinking about politics and the state

    So, I've made this request before and I'd like to make it again. Can you please consider writing a book on how an ordinary citizen can go about thinking about policy in a liberal republic? What does it mean to aspire towards a liberal republic, when it comes down to choosing the right policies. And so on. The writing is already done in the sense that it could just be a compilation of blog articles on the topic of public policy, like the ones above.

    I think it would be invaluable. It would do a world of good to the significant percentage of young folks in the country.


    1. Dear VS,

      Thank you for the kind words. A book is in the works! Hopefully will come out in early 2019. It is a fully integrated and internally coherent book, not just a collection of blog articles. But the blog and the book are both products of the same Weltanschauung and you will hence see the book as familiar territory.

    2. That is awesome! However, I wonder if it will be heavy on wonkery, catered to the more engaged, policy-educated audience or whether it will be for the ordinary, young citizen and more philosophical if you will. I think both would be valuable ofcourse but I was trying to make a case for the latter.

      Regardless, I will be looking forward to the release eagerly. Thanks.

      I also had in mind something like Jug Suraiya's collections of his newspaper essays. The benefit being that the publishing can be done quickly, using the specific issues that the public is concerned about at the time as examples, thereby increasing the chances that the ordinary public engages and picks up the book in the first place. And considers it as light piecemeal reading, and gets into the philosophy in steps.

      Like one would over time on subscribing to your blog, one article at a time...

  3. An increase in fare by 25% last year led to a decline in ridership by 15%. This not only results in loss of revenue for the metro authorities but also adds more stress to the public road system leading to calls for more investment in flyovers and upgradation of existing roads. Therefore, higher user charges do not by any means seem to be an effective solution. Comparing different metros may be good, but in does little to provide effective solutions. A better way would be to look at the cost of upgrading Delhi's metro vis a vis other alternatives in the city.

  4. > An increase in fare by 25% last year led to a decline in ridership by 15%

    Do we know that x caused y?

    I would have expected highly inelastic demand. Once a person starts using a metro, the price does not matter much.

    1. Delhi metro had seen average daily ridership figures go up by at least 2 lakh/day year for the 8 years before 2018. So the fare hike is a very likely candidate for the drop in ridership.

      In general, a major shortcoming in this analysis is that you have not taken into account the trend of Delhi metro ridership. It has increased by 2 lakh/day almost every year. That rate would put it on track for reaching the 60% rise in about a decade's time.

    2. Agreed. We could certainly improve our analysis on Delhi Metro. Data availability becomes a problem for a deeper analysis. If you could point us to some sources we could try to expand the analysis. We would also welcome any analysis that you have done. This piece concentrated on comparing Delhi with other large metro systems. A trend analysis of Delhi metro can be a follow up piece.

  5. The comparison I would make is with metro systems that were built in the same time frame as the Delhi metro, and currently have similar network sizes. From the wikipedia list, these are,

    Chongqing (743.1 million, 266.2 km, 17 million)
    Nanjing (977 million, 378 km, 12 million)
    Shenzhen (1654 million, 286.2 km, 23 million)
    Wuhan (927 million, 288 km, 19 million)
    Delhi (1008 million, 317 km, 26 million)
    Tehran (717 million, 173 km, 15 million)
    Guangzhou (2800 million, 391.8 km, 25 million)

    We can see that the cities where metros see extremely high traffic are coastal economic centres (Guangdong and Shenzhen) which tend to have extremely high population densities.

    In its comparator group, interior cities with large populations (Wuhan, Chongqing) Delhi comes out looking pretty good. Nanjing, a slightly smaller city, has good numbers for its population but has a longer network.

  6. Beijing, Mexico City and Paris are not ports but they seem to transport more people per km. There is evidence to suggest China may have over-invested in infrastructure.


Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticism is perfectly okay; uncivilised language is not. We delete any comment which is spam, has personal attacks against anyone, or uses foul language. We delete any comment which does not contribute to the intellectual discussion about the blog article in question.

LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.