Some thoughts on metro railway systems in India

This post will analyse metro railway projects in India, highlight their failings and make some suggestions.

There are three reasons why metro projects globally struggle to be sustainable, even on the Opex. One, the original traffic projections were on the higher side, making the revenue forecasts a non-starter. In other words, the metro should never have been built. Second, the ticket prices were kept too high, resulting in less demand. Third, there is limited induced ridership due to the lack of TOD, parking facilities, and modal integration. 

The first is a common factor associated with projects globally (optimistic forecasts to justify otherwise unviable projects). The second is a tricky problem insofar as lowering it further risks significantly worsening the financials of the metro. Besides, there is enough evidence that metro ticket prices are reasonable across cities. And the real problem is less the fare level than the generalised cost of the trip, or the fare plus the last-mile penalty (arising from the need to change modes, and the lack of any modal integration).

The last one is especially important since the metro is a large investment that is made to shape the city's growth, specifically by densifying the walkable vicinity of its stations and thereby inducing captive ridership. This is the biggest source of sustainable passenger ridership for the metro. Thin station catchments may well be the binding constraint to the operational viability of India’s metro systems. 

India's experience with TOD has been disappointing. None of its 900-odd metro stations is a TOD (densified residential and/or office spaces, which enable walkability from the station to house/workplace). This is unsurprising since the TOD policies generally suffer from poor design. There is also the reality that very few metro projects globally can recover even opex with the farebox collections and rely on real estate and other commercial revenue streams. More on it later. 

In addition to the three demand or revenue-side challenges, there is also a cost-side problem. They include the wrong mode selection - choosing an underground full metro when light rail or bus rapid transit system would have sufficed. The low-cost foreign loan is deceptive since a depreciating rupee inflates the cost (though the Yen, the major source of metro debt, has been depreciating or holding steady against the rupee). Further, the linear stunted network topology, as against a connected grid, limits ridership. Most Indian systems are linear stubs, so they sit permanently below the ridership threshold at which farebox economics turn. As an illustration, Chennai’s finances turned around dramatically once Phase 1 closed the loop, with ridership surging from 13% of projection at CAG FY20 to now ~60%. 

I asked Claude to generate certain parameters for the leading Indian metros. First, the difference between the DPR ridership projection and the actual realised ridership points to strategic misrepresentation with inflated estimates to secure project approval. As can be seen, apart from Delhi, the shortfall is very high, with the majority not even able to generate a third of the estimated ridership. This points to clear strategic misrepresentation to get the project sanctioned by exaggerating the ridership estimates. 

Delhi is now above its old 2019-20 target after Phase-III/IV and post-COVID recovery. This points to the fact that maturation and network completion, and not fare tweaks, are likely to move the needle.

In terms of the percentage of just the opex covered by farebox revenues (or ticketing), it is very low in many metros. However, when combined with the other revenues, five metros are able to cover the opex.

But if we include debt-service, the picture becomes extremely bleak. Most Indian metros are deeply in the red. 

This has important fiscal implications. I took six major states, and assumed their current ridership trajectory and fare growth to calculate the gap between their total revenues (farebox and non-farebox) and total opex and debt service. This assumption overlooks that ridership trajectory may improve over time. In any case, this analysis points to an undiscounted 25-year total gap of ₹1,06,569 cr, or NPV of ₹64,423 cr when discounted at 7%. The state share is around 60% or ₹38,582 cr, and the central share is ₹25,840 cr, amounting to a total annual budget requirement of ₹6,900 cr for FY26 to ₹6,256 Cr for FY30. 

A global comparison yields interesting results. Apart from Northeast Asia, generally, metro railway systems struggle to cover even Opex through farebox collections. 

In all western cities, their governments funded the infrastructure (versement mobilité in Paris, LTA asset ownership in Singapore, TfL/Treasury grant in London, MTA authority-level bonds backed by dedicated taxes in New York). The operator, therefore, never sees the debt service, which has been socialised as a public subsidy up front. Only Hong Kong MTR and Tokyo make the operator carry capital and still clear cost, and they do it through property and ancillary income, not fares.

India is a structural outlier in loading the capital debt onto the SPV's own balance sheet. The Indian farebox is being judged against a burden every comparable peer either socialises to the budget or funds from property. The sustainable systems didn't find a magic farebox - they either kept capital off the operator (the European/American subsidy model) or monetised the land (the Hong Kong/Tokyo model). India did neither.

So what can be done?

A systemic reform requirement is to orient the design and implementation of metro railway systems away from being primarily a railway construction project to an urban planning initiative. This would entail tight integration of the planning of metro rail projects by the municipal authorities. 

The TOD influence zone (TIZ) around each station should be seen as a genuine opportunity for redevelopment and densification. The planning regulations applicable to the TIZ should make it significantly more attractive to build insidecompared to outside - 3-4 times higher FAR, remove parking minimum, TDR-driven land assembly, TDRs brought in with higher multiples, etc. Equally important, the regulations should make it attractive to redevelop the built-up areas (and not be like this). Instead of being prescriptive on land use, it should be left to market forces to determine the appropriate shares of different types of uses in each zone. It would also allow the markets to crowd in the right kinds of commercial amenities, such as groceries, retail, restaurants and cafes, and services, into the zone, depending on its broader land-use mix. 

An important reason for the failure of TOD is the lack of attention given to its public realm elements, specifically the street configuration for connectivity and walkability, and public amenities. This is a bad omission given that walkability (from/to commute and for amenities) is the primary objective of TOD zones. An area without these elements contradicts TOD. Unfortunately, in built-up areas (which form the major part of TIZ), these elements don’t emerge overnight. They require master plans that are configured to enable walkable street connectivity, higher density, and mixed-use developments. Once the master plan is in place, its implementation should be facilitated proactively through the development control regulations and building bye-laws that allow for TIZ preferences and incentives, and instruments like the Town Planning Schemes. These are 10- to 15-year agendas.

The TIZ preferences should be complemented with feeder buses, multi-modal fare integration, and e-rickshaw access so as to extend the catchment beyond the TIZ. Further, as the Chennai example shows, metro lines should complete the loop or connect the grid, and not remain as linear stubs. 

For financial sustainability, the TIZ should deploy appropriate value capture finance instruments. I blogged about it here. They would include betterment levies (brownfield areas) and impact fees (greenfield areas), station-area commercial development, air-rights leasing, and purchaseable FAR. However, care should be taken to ensure that these levies do not detract from the commercial attractiveness of building inside the TIZ. These revenues should be assigned to the SPV balance sheet. This is how HK MTR earned 65% of its 2024 profit from property, not fares. 

These urban planning and other policy requirements should be tightly incorporated into the approvals given for these projects. Every new metro DPR must include statutory master plan amendments, FAR densification for station catchments, mixed-use zoning, land assembly strategy, and the land-value-capture financing plan. They should be preconditions, not annexures. The rail alignment should be a subset of the urban plan. 

There should be no sanction without station-area FAR increases, TDR-driven consolidation, betterment-levy design, and air-rights leasing, all routed to the SPV. TDR can pre-fund land assembly before construction starts. These should become an approval veto. Further, any ongoing support from the Government of India must be made contingent on compliance in letter and spirit with these requirements to pre-empt backsliding and subversion by the states. 

The planning elements discussed above would have to go against the conventional wisdom on urban planning held close to heart by town planners across India and the vested interests of builders and others. It would therefore require strong resolve at the highest political and bureaucratic levels to overrule objections and enforce them. 

For new projects, central funding should become conditional on a corridor-density threshold, independently audited ridership models, and ex-post accountability. There should be an end to the practice of a self-certified 14% IRR. Anything below the threshold should qualify only for a MetroLite / MetroNeo / BRT.

In conclusion, the urban metro should be used as an opportunity to redraw the metro as an urban plan by densifying station catchments, capturing the value, and servicing the construction capital.

The demand side constraint of the Indian economy

I blogged here, arguing that Indian businesses face a structural cost constraint similar to that of developed economies that erode their global competitiveness. This cost constraint on the supply side must be seen alongside a constraint of a consumption class that is far smaller and shallower than imagined. In other words, a developed country's cost structure in a low-income country's market size. 

This point has been a consistent argument of this blog and was documented in great detail in Can India Grow? in 2016, much before it became more widely recognised.

India’s demand-side problem is twofold. For a start, it already has a low per capita income, the lowest among its emerging market peers. Second, this itself conceals a low base of the consumption class that is also not widening proportionately with economic growth. The nature of economic growth is such that it is not only not broad-based enough but is also widening inequality. 

The recently released PRICE - Tata Sons Many Urban Indias report is a confirmation of the demand-side constraint. It finds that the top 100 cities, with just under a fifth of the population, capture almost half of all household surplus, the closest measure to genuine discretionary spending capacity. What sits outside, the other 81% of India, holds only about 53% of surplus combined, and most of that is in the top slices of tier-2/3 towns and the productive rural belt. The surplus-generating layer of the economy is essentially urban and already fully contained. There is no large pool of "middle India" outside urban India waiting to broaden the base.

PRICE's top-100 city-tier framework (Big Six, Boomtowns, Breakout, Frontier) is the empirical proof that "urban India" is not a homogeneous consuming class. The gap in savings between the top and bottom tiers is nearly 3.5×, and one in six Frontier-city households is already financially overstretched. This is what “narrow” looks like at the city level. 

The top 15 cities have two-thirds of all top-100 consumption, and the Delhi NCR alone has 15% of all urban consumption. The consuming class is a stratum within the Big Six, not the Big Six themselves. Even within the six megacities, PRICE reports that 24% of households are low-income. The consuming class is not "urban India" (~500M) or even "top-100 city India" (282M), but a stratum of prosperous households scattered across the Big Six and the top of the Boomtown tier. 

The Blume Ventures categorisation of India’s consumption class finds three groups, with the first comparable in size to the population of Mexico at 30 million households (the Blume Rule of 30, or 10% of Indian households). A worrying finding is that the consuming class that anchors private demand is small and getting deeper, not significantly wider. A vast base has almost no discretionary spending power at all. This is the inverse of the broad, upwardly mobile middle that sustains private investment.

The PRICE report's optimistic headline (~715M "middle income" people by 2031) conflates a household that shares a scooter with one that services a car EMI. Once you apply a real surplus filter - savings above ₹4 lakh a year, which is roughly what it takes to run a car, insure a family, and take one flight - the number lands where Blume said it did. In simple terms, PRICE's 715M by 2031 measures how many Indians will earn enough not to be poor, a significant achievement. Blume's 140M measures how many can act as a consuming class today - the market a brand can actually sell a car, an AC, a holiday abroad or a health-insurance policy to. 

The numbers are further validated by other sources. An income assessment by Rama Bijapurkar points to a more differentiated group, but a similar narrow base of the consumption class. She estimates that 93% of households have annual consumption of less than $5,700. 

The data on penetration of discretionary goods and behaviours corroborates the Blume survey data. Almost every independent marker lands near the same Blume Rule of 30. Whether you measure cars, ACs, credit cards, foreign travel, or who actually pays income tax, the consuming class converges on the same narrow apex. If anything, the Blume data itself looks too optimistic. 

Whichever marker you pick - cars (8%), ACs (13%), passports (~10%), income-tax payers (1.5–3%), credit cards (4.6%), foreign trips (2%) - the consuming class lands at ~10% of households or fewer. That is also exactly where the bottom-up Blume spending model, income surveys of Bijapurkar, and asset penetration estimates of NFHS/Bain lands.

India's "middle class" has been estimated at 29 million people one way and 600 million another, a multiple spread of 20 that reflects what one is choosing to count. 

India has a consuming class of ~140M, an aspirant class of ~300M moving toward it, and a base of ~1bn for whom the "middle class" debate is moot. That picture is consistent with every methodology once you ask which question each one answers - how many can afford a car (~10%), how many are above poverty by global standards (~30%), how many participate in the digital economy (~65%). Conflating them is the single most common error in India market-sizing. This is a very nice summary of why nobody gets the elephant in full.

In this context, it is also useful to draw a nuance on India’s high consumption rate of 61% of GDP, which is far higher than China’s ~38% of GDP. India’s problem is that while it is already a consumption-led economy, the problem is the quality of that demand. It is too thin per person, skewed to the top decile, and increasingly financed by debt as households run down savings. In other words, the demand side constraint is not the consumption share of GDP, but that there are too few middle-income earners to generate broad, income-financed volume. That points policy at jobs, wages and investment, not at “stimulating consumption.”

Fundamentally, all the above is a reflection of two important factors - the low per-capita income and the widening inequality in sharing the benefits of aggregate growth. 

India's per-capita output (PPP) trails Vietnam and Indonesia and is under half of China's. Its Mexico-like consuming tier is real, but at best, only ~10% of the country.

India imposes rich-country costs on poor-country incomes in capital, fuel and land - and because those costs are policy-made, they are the actionable half of the problem. The demand blade can only be widened the slow way: through jobs, wages and productivity.

Unfortunately, corporate surpluses are being captured, not recycled into the wages that would create the next tier of consumers. Profits race ahead while sales, jobs and pay lag far behind.

It also does not help that Indian businesses across sectors are averse to spending on R&D and innovation. 

R&D is a fixed cost that only pays off over a large, contestable market. A thin domestic market depresses the return to it, which is why the economies that lead on R&D expenditures are either big-and-rich at home (the US) or export-driven (Korea, Germany, Taiwan). India is neither, and sits alone in the low-R&D corner.

This is an instructive comparison of India with China and South Korea.

The main takeaway is that India’s existing consumption class, confined to just 10% of the households, is too narrow to sustain high growth rates for long periods. It requires considerable broadening. But broad basing economic growth requires access to good jobs, a daunting task given that the gig economy is the biggest source of job creation and elsewhere it is mainly contractual jobs. Worsening matters is the reluctance of the private sector to invest or innovate. 

Weekend reading links

1. SpaceX IPO is ample proof that the Chinese wall between equity research and investment banking in IB firms is a myth.

You might expect wildly divergent views on a company as speculative as SpaceX. Here the underwriters disagreed only over the scale of the upside. Sceptical voices came from outside the syndicate. Morningstar, for example, valued the shares at $63.

This is now dead

When the dotcom bubble burst, regulators uncovered emails showing that Wall Street analysts were privately disparaging stocks they were publicly touting. A 2003 global settlement between banks and regulators on analyst research imposed sweeping restrictions. It barred investment bankers from influencing analyst compensation, tightly controlled communications between research and banking, and banned analysts from IPO pitches and roadshows. New York Attorney General Elliot Spitzer’s premise was that shielding analysts from bankers would deliver truly independent — and better — research. On one level, the reforms succeeded. Banks have constructed a robust compliance apparatus to wall off research from investment banking. “Chaperones” now police interactions between analysts and corporate finance to prevent even the appearance of pressure. The Spitzer global settlement formally ended last December in favour of more flexible rules overseen by an industry association.

It is hard to see how SpaceX shares can avoid a cratering.

About 30 per cent of the roughly 640mn SpaceX shares available to trade have been borrowed to sell short, up 10 percentage points over the past 10 days, highlighting how traders are becoming increasingly sceptical of the company’s market prospects... About 900mn further SpaceX shares could become available to trade as soon as next month when certain lock-up provisions for pre-IPO investors expire. Traders who doubt there is sufficient demand for the deluge of extra equity are cashing out now as a result, market participants say.

2. After reducing their hiring last year, firms with jobs exposed to AI are planning to increase their entry-level hiring this year, but they come with a "seniorisation". 

Candidates for starter roles in the most AI-exposed industries are now expected to show a mastery of the skills traditionally demanded of more seasoned staff, such as data-driven decision-making and people management... Eleanor Lightbody, CEO of Luminance, which develops AI for the legal profession, says these middle layers could be squeezed out altogether, “because we are going to hire more juniors, who are really going to understand how AI works, and more seniors [are] staying in the business because [by using AI], they can be more productive and have more capacity”. The challenge for job seekers is that it is hard to find entry-level jobs that allow them to develop the higher proficiency that seniorised roles now require... Reliance on AI is increasing the demand for distinctively human “soft skills” — or “power skills”, as some are now calling them — such as creativity, empathy, judgment and networking ability. PwC’s jobs report found that new tasks added to job adverts for AI-exposed roles were two and a half times as likely to call for such capabilities.

3. Very good description of how China became so dominant.

It enticed unsuspecting giants such as Apple, Tesla, Motorola, and Lucent with low-cost logic. When sufficient local manpower was trained, subcontractors developed, and stakes became important, China applied the squeeze. China would break contracts, cancel licences, coerce the transfer of technology, control pricing, withdraw incentives, force equity participation, conscript technology and evict them. Huawei, BYD, CATL, and SAIC are some examples of the resulting indigenous giants that emerged. In an act of silent invasion, conscripted technologies have been converted into military capability. It is dominant as a supplier of several raw materials, such as rare earth minerals, gallium, graphite, and lithium; a dominant buyer of soya from Brazil and iron ore and wines from Australia; a financier of BRI projects; and a provider of processing technologies for Chilean copper and lithium in select South American countries. With this web of dependencies, it can choke several factories.

4. Janan Ganesh feels that for Britain to start making real reforms, the incoming PM Andy Burnham must do more welfare and subsidies and discredit the whole .ideology.

What is the precedent for a rich democracy doing pre-emptive economic reform? Which nation ever made controversial structural changes — involving winners and losers — to prevent a crisis, rather than in response to one? Southern Europe needed the Eurozone panic of 2009 onwards to make spending cuts. Hawke, Keating, Margaret Thatcher and Ronald Reagan were reacting to 1970s stagflation. François Mitterrand in 1983 was reacting to a market shock that to some extent he’d created. Reform only happens when it absolutely has to happen. So try again, prime minister. Fail again. Fail worse.

5. Japan embraces a more proactive government-driven economic growth policy.

Sanae Takaichi's cabinet approved a policy blueprint that targets a combined $2.3tn of public and private sector investment between now and 2040 in 17 chosen sectors. Ministries will be able to make budget requests without upper limits; budget construction, according to the document, will be “fundamentally” changed. Much of the blueprint is about economic security, but a refreshingly large amount is about growth... possibly the most meaningful lines in the new strategy place Japan’s future efforts in the global context. “Among advanced countries, there is a big trend of the government and private sectors working together on large-scale, long-term industrial spending,” it read... The government would strive, the document further promised, to meet the challenges of “the era of great global competition between industrial policies”... 

On the same day that the blueprint was agreed, the Ministry of Economy, Trade and Industry produced separate guidance for growth investment — an effort to encourage Japan’s 4,000-odd listed companies to shift more of their endeavours towards growth and a witheringly blunt critique of how matters are at the moment. Within Japan’s 350 largest companies (by sales), 65 per cent of invested capital remains locked in what it calls value-destructive segments, according to METI’s research. In the US, the equivalent ratio is 39 per cent. Both Takaichi’s blueprint and the new METI guidelines are attempting a new version of industrial policy that not only seeks to spur growth, but places a huge bet on the government’s ability to encourage companies in a way that market forces have not.

6. Kevin Warsh is trying to scale down forward guidance

“Financial market prices are probably the most important source of information to guide central bankers,” he said at his inaugural press conference last month. “But when all the financial markets are doing is reflecting back what we’ve said, then we’re taking the most important source of information and we’re being blind to it.” Many agree with the Fed chair’s view that central banks’ focus on predictability has led to a world in which markets obsess more over what officials say than what is actually happening in the economy. “Forward guidance has turned markets into a mirror,” says Ajay Rajadhyaksha, global chair of research at Barclays. “The Fed watches markets; markets watch the Fed. And no one’s actually watching the economy.”...

Recent research by the US central bank suggests its decisions also have an outsized impact on equity markets, leading to big changes in how investors price stocks. Advocates of forward guidance say it helps avoid the sort of surprises for the markets that feed overall volatility and eventually raise borrowing costs as investors demand greater compensation to stomach market swings. Warsh and his allies counter that the attempt to pacify markets simply encourages greater risk-taking, damping short-term volatility but storing up bigger shocks for later. The debate is all the more important because of the backdrop: the surge in borrowing over almost two decades. Government debt around the world has risen sharply following the global financial crisis, the Eurozone crisis, the pandemic and the wars of the 2020s...

Some analysts highlight the so-called taper tantrum of 2013 — when markets were unnerved by Fed statements about its plans to shrink its balance sheet as a result of officials’ false sense of certainty. Others link the 2023 collapse of the US’s Silicon Valley Bank to central bankers’ previous pledge to keep interest rates low.

This is a very important factoid about Fed communications.

Between 1990 and 2022, an era in which the status and prominence of central banks steadily grew, US 10-year government bond yields fell by more than 7 percentage points. All of the downward moves throughout that period took place during the three days around Fed meetings, rather than in response to political events or economic data. But the relationship broke down after the Fed appeared flat-footed on inflation in 2022. The Riksbank data shows that the central bank’s meetings had little to do with the subsequent rise in yields.

An important concern for the bond markets is the sharply increased volume of Treasuries held by hedge funds that use leverage to bet on tiny differences in interest rates. 

The Fed estimates that large hedge funds’ holdings of US Treasuries doubled between 2023 and 2025, faster than the growth of the market as a whole. Such funds now own more than $2.5tn in Treasuries, according to the US central bank and US Treasury data. Such sums dwarf China’s official holdings — not an exhaustive account of the country’s exposure — which have fallen from $1tn in December 2021 to $659bn in May 2026, according to US Treasury statistics. Moreover, the Bank for International Settlements, the central bankers’ bank, has warned that the hedge funds might have to dial back their stakes in government bond markets even quicker than they arrived — a possibility it describes as one of the most troubling financial stability risks in the world today.

Then there are the other concerns for bond markets.

The Bank of England said this month that AI hyperscalers borrowed more in the first half of 2026 than in the whole of 2025. So far this year they have accumulated as much new debt as the UK government. An interest-rate surprise from the Warsh Fed might not only unsettle Treasury yields but set off a vicious circle of margin calls — when a sudden price movement in assets bought with borrowed money requires an injection of capital — and fire sales by hedge funds that could create a dash for cash.

7. A cautionary tale on the difficulty of private enterprise establishing and managing entire railway systems comes from the example of Brightline Express, which connects Miami and Orlando, and became operational in 2023.

Brightline traces its roots to 2007, when Edens’ Fortress spent $3.5bn to acquire Florida East Coast Railway, then a listed freight transport company. Its tracks spanned from north to south in the Sunshine State, and construction for what would become Brightline began in 2014, with the Orlando service beginning a decade later. In a 2024 bond prospectus, Brightline executives forecast that by 2026 they would have nearly 8mn annual riders, split roughly evenly between the Orlando-Miami route and a more local service in South Florida. That base of customers was expected to generate $700mn in revenue... Even with customer levels up 16 per cent year to date through May this year, ridership will struggle to hit 4mn this year. Operating income has at best reached break-even before debt service costs... 

The promoter Wes Edens’ Fortress wrote off its investment long ago, and it is now hedge fund bondholders who are jockeying for control of Brightline. But alongside distressed debt specialists such as Nut Tree Capital Management, Aristeia Capital and Redwood Capital Management, there are a handful of more staid asset managers including Nuveen and First Eagle also at the table. As well as corporate bonds issued by subsidiaries, the group’s debt stack includes more than $2bn of traditional municipal bonds, half of which are guaranteed by a bond insurer.