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.

The cost of doing business in India - land prices

I blogged here, highlighting the high cost structure that Indian businesses face. This post dives deeper and examines land cost.

A comparison with peers shows that urban land is priced like that of a rich country and ends up hoarding the nation's savings, with 77% of the households parking their wealth in real estate. Mumbai ranks among the world's 20 most expensive prime markets. Industrial land across the major Indian cities is prohibitively expensive compared to peers.

An Indian manufacturer pays roughly 5–15 times more for industrial land than a Chinese counterpart in a comparable tier-2 city and 1–4 times a Vietnamese peer. 

I used Claude to generate the graphic below on the total land transaction cost across states, combining the stamp duty and registration fees and the change of land use fees for the conversion of agricultural land. While stamp duty has converged to 6-7% of land value, the total cost varies widely across states. Interestingly, the southern states have the highest transaction rates. 

Four states can’t be reduced to a clean number - Maharashtra charges a premium of 25–100% of the ready-reckoner differential; Gujarat charges on the agri/non-agri differential; West Bengal rarely grants conversion at all; and Delhi’s urban land is already non-agricultural. For these, the combined figure shown is a floor, not a total (hence the ‘+’). Maharashtra, in particular, would likely top the chart if its premium were expressible as a share of land value.

Andhra Pradesh (deemed conversion since 2018, NALA Act now being scrapped) and Uttarakhand (permission eliminated for specified categories) are the only states that have removed conversion permission as a bureaucratic step. Assam, J&K, Odisha, Puducherry and Tripura have moved to negative lists.

I then asked Claude to generate the typical cost of industrial land in some representative locations across India. The wide variations in the construction levies as a share of land value arise from the variations in the underlying land valuations (the % is higher in places with lower land values). While the state and local taxes make up 10-17% of the land value, all the land and building-related taxes make up a near-uniform ~10% of the industrial project cost (land, construction, taxes).

I did the same for residential land cost at representative locations in a few cities. Ignoring the Gurgaon outlier, the state and local taxes make up 16-26% of the land value, and all the land and building-related taxes make up 18-25% of the price paid by the homebuyer. In other words, the tax burden is a fifth to a quarter of what a homebuyer pays.

In both cases, the lower land valuations in places like Lucknow or Indore or Vizag mean that the share of the nationwide levies (GST and labour cess) is much higher. In other words, these levies bite more in the second-tier and below cities and end up eroding their comparative advantage.

A like-for-like global comparison for industrial land is difficult given the lack of data. The graphic below is a broader international comparison, using the World Bank’s last Doing Business 2020 survey, for the construction of a typical (2 storeys, 1,300 sq m, on a 929 sq m plot, valued at 50× GNI per capita) warehouse (as a % of the warehouse value) in each country’s largest business city. It measures every statutory licence, permit, inspection, utility connection, non-recurring tax and property registration fee, but importantly, excludes VAT, capital gains and the price of the land itself.

India’s statutory fee stack is middling, sitting below Mexico, Indonesia, Bangladesh, and Delhi at 2.8% is not too far from the OECD high-income benchmark of 1.5%. This questions the reasoning that Indian construction-permit fees alone make land prices uncompetitive. However, India’s manufacturing competitors like Vietnam, Thailand, Brazil, and Poland levy less than a fifth of what India does on the same measures. 

More importantly, the DB survey data excludes land price, stamp duty, conversion charges, and the GST input credit, precisely the costs that were the basis for the discussion earlier. The state’s tax share at 18-43% of industrial land cost, or 9-11% of total project cost is where the land-related competitiveness gap sits, and not in the counter fees that the WB’s DB Survey tracked.