But for many non-resident Indians (NRIs), there is something different.

Recently, I was talking to a colleague who’d just wrapped up a European trip. They were overjoyed to be back in a place with an AQI of 130 from 9. During our conversation, I asked them what they missed the most. And what disappointed them about Europe?

The answer was kind of refreshing; I am not gonna lie. I was expecting the third point they made to be the primary one, like most do. House help & 10-min delivery.

This is something that has been sitting in an uncomfortable middle for me for a long time. I get the point about the double thank-you moment. You are paying someone for their services. You both are better because of that. The househelp is getting paid, otherwise they wouldn’t be able to. And you are not getting bothered about the chores, and can focus your energy somewhere else.

The other side is this: this is not a glamorous job at any cost. And this is a big bug of the Indian socio-economic scene. This shouldn’t be celebrated as a big feature that we often tend to do. Yes, they are doing better because of the opportunity, because I am paying. But they could’ve done far better if there was a better opportunity.

Anyways. I thought of quantifying my discomfort. And I went down a deep rabbit hole. I have spent a lot of time and tokens on this, more than I would have liked to. This probably needs multiple blog posts to do a proper narrative-driven summary.

In any case, it would be good to summarise a few things.

So the actual question. Is domestic help in India cheap? And Cheap compared to what?

Subscribe now

Drawing a line in the sand is often important in analytics. So here is one quick assumption that I made: India’s PLFS data breaks down what Indians earn on average: self-employed workers make about ₹13,279 a month, regular salaried employees about ₹20,702, casual labourers about ₹12,750. I made a weighted average of how much of the workforce falls into each bucket (roughly ~ 56%, 24%, and 20%), and it turned out about ₹15,000 a month.

But that’s everyone. Street vendors, casual construction labour, salaried clerks- everyone is included there. Domestic help is a much narrower and more specific market. Not everyone can buy it, and it’s concentrated in T1 and T2 cities. Ask around in a Mega city, and the number you’ll actually hear for full-time domestic help is somewhere between ₹20,000 and ₹30,000 a month. So that’s what I’m using here: ₹25,000, the midpoint.

The obvious way to check if this is “cheap”: price a full-time domestic worker against the country’s own GDP per capita. And check if you can spot a pattern there.

And the naive answer surprised me. India isn’t the cheapest country on this list. It’s the most expensive, by a wide margin. A full-time domestic worker here costs more than the average Indian’s entire annual income. Not a large share of it. Even if we use the earlier discussed 15k number, the percentage would still be close to or more than 100%.

Nowhere else in the sample gets anywhere close, rich or poor. This holds almost everywhere where labour is priced locally, whether the country is poor or comfortably rich. It only breaks for a handful of places (Singapore, the Gulf, Hong Kong) where they don’t price domestic help off their own wages at all but rather import it from somewhere cheaper. Another country.

So the popular intuition, that help is dirt cheap here because we’re a poor country, goes for a toss if we take a look at GDP per capita numbers.

But comparing cost to GDP per capita has a slight problem. GDP per capita isn’t what a person actually earns. It’s the country’s total economic output divided by literally everyone, including your niece who’s still in school. There’s a sharper, more honest question hiding underneath the naive scatter plot.

How many hours of someone else’s work can an hour of your own work buy?

This is where it stopped being an interesting fact and started being a bit uncomfortable. Everywhere else in the sample, an hour of average work buys more than an hour of domestic help. In India, it buys less. An average Indian’s hour of work buys less than an hour of a domestic worker’s time, not more. (Singapore’s number is enormous, but for a completely different reason. That’s imported labour, as we have already discussed.)

Also - one thing to keep in mind here. Which is the crux of the matter. We already have established that domestic help is not a glorious job. But for a large section of the Indian social strata, even the opportunity of doing that in a T1 or T2 can feel aspirational.

So that’s the number underneath my discomfort, finally. It’s not that help is a bargain here. It’s that too many people don’t have a better option than to sell an hour of their time for less than it’s worth. This is, for sure, not a feature of how efficiently the Indian state has organised the economy - and that probably shouldn’t be celebrated.

This is just what a very low floor looks like, once you measure it. And the low floor might be lucrative for many.

While writing this post, I randomly remembered something an extremely annoying (he was complaining too much about Kolkata, his ex-wife, parents and many other things) Uber driver once told me: how can you tell if someone is really rich? Just see if they can hire a full-time driver.

Observations, data, and the not-so-ordinary world. Subscribe if that's your kind of thing.

Subscribe now

Read more of my writings here, and learn about me here. I also write essays on tech, product, and strategy over here.

If you found this interesting, I would love to hear your thoughts.

Notes on the data

Chart 1 — Domestic help cost vs. GDP per capita (20 countries)

GDP per capita: IMF WEO, nominal 2025, all countries.

Domestic help cost (% of GDP per capita), by source confidence:

Stated assumption: India, ₹25,000/month, full-time domestic help in T1/T2 cities.

High confidence, real statutory/government figures: Philippines (DOLE Kasambahay wage order, NCR minimum), South Africa (Sectoral Determination 7 / BCEA minimum wage, R28.79/hr), Hong Kong (Migrant Domestic Helper Minimum Allowable Wage, HK Immigration Department).

Medium confidence, aggregator or agency-rate based: China (city-level ayi/cleaner rate estimates), Indonesia (Jakarta live-in ART estimates), Thailand (Bangkok live-in estimates), Mexico (empleada de planta wage estimates), Brazil (mensalista wage estimates), United States (full-time housekeeper/nanny salary aggregators), United Kingdom (housekeeper salary aggregators), Germany (Putzperle/go-quitt minijob-rate sources), Singapore (FDW wage, MOM-adjacent sources).

Low confidence, thin sourcing, treat as plus or minus 10 points: Vietnam, Turkey, Nigeria, Egypt (live-in domestic wage estimates from limited sources), Qatar, Saudi Arabia, UAE (migrant domestic worker wage, agency-adjacent sources).

Chart 2 — Hours of domestic help one hour of average work buys (8 countries)

Singapore’s ratio (8.46) reflects imported labour priced below Singapore’s own wages, not local wage compression. It’s structurally different from every other row in this table and shouldn’t be read as “cheap” in the same sense India, Brazil, or Germany are.

The weakest links in this table, flagged plainly: Brazil’s and South Africa’s average-wage figures, and the UK’s, don’t have a primary-source citation behind them. Everything else traces to a named statistical office or government wage order.

I didn’t pay much attention to exit polls, and what they had projected, until someone brought to my attention one tweet which was supposed to be “exposing” the fraud techniques of PPA.

At that moment I found it quite interesting 1. I tweeted about it but couldn’t write this one up. I had some major family functions. So I thought I would write up why I found their numbers and methods quite interesting. Being an Analytics professional, it made some sense to me.

First things first, one thing that we need to keep in mind: every model is wrong. Some are useful. There are some assumptions that you need to make while modelling.

In this video, we can hear the head of PPA talking about their assumptions and methodologies. They apparently covered more than 6 lakh voters in West Bengal. 4.5 lakh were male and 1.5 lakh were female. They started doing this 4-5 months before the poll. They seemed to be following a bottoms-up approach, as opposed to Axis My India which follows a top-down approach. Pradeep Gupta of Axis My India had mentioned this distinction2. Their first job is to gauge who is winning the election (a macro trend), and then they try to predict how much everyone is getting. PPA seems to be doing the opposite.

Anyways, doubling down on their methodology now, one by one.

4.5L Male vs 1.5L Female

I had an assignment in my previous company. I had to build a model to predict whether a doctor would recommend a prescription drug. We found that the presence of one particular risk factor almost guaranteed that this medicine would be prescribed, roughly 90% of the time. If it was absent, things became trickier, and we had to do some complicated modelling for that.

What we did was simple. If the risk factor is present, it is automatically a positive case for prescription. For the rest, a model is used.

When someone pointed out the 4.5L vs 1.5L split in a reply to that tweet, they called it a biased sample. I thought the opposite. This imbalance is a feature, not a bug.

Women in West Bengal have overwhelmingly voted for TMC. Maybe except a small section, the preference is near-deterministic. So once you start calculating from your sample and see that irrespective of age, location, or anything else, a statistically significant number of women voters are going to one party, you don’t really need more of them. A homogeneous group doesn’t need a large sample to be estimated accurately. You can simulate voters in each constituency from a smaller base, and make a healthy projection.

Male voters are the ones who differ a lot demographically (at least that has been the case historically). Age, area, community, occupation: all of it plays a role. For this, you need to make many cuts of the data and find statistical significance at each level. So it absolutely makes sense to have more people from that category.

Is the number enough?

With 6 lakh total across 294 constituencies, you are working with roughly 2,000 effective samples per constituency on average. The margin of error at 95% confidence works out to about plus or minus 2.2%, and with a realistic design effect accounting for clustered sampling, you are closer to plus or minus 3.5%.

The male sample (4.5 lakh across 294 seats, roughly 1,530 per constituency) is where the real modelling happens. And here is where it gets a bit uncomfortable. PPA is doing bottoms-up. That means they are building the prediction seat by seat, constituency by constituency, not detecting a broad state-level wave and working downward. For that approach, 1,530 respondents per seat with a margin of error of 3.5% is genuinely thin. West Bengal has had plenty of seats decided within that margin. So unless, you get a overwhelmingly large signal with male + female combined, it is a bit difficult to give a verdict for a seat.

So the sample size is the honest weak point (but apparently- this is still the highest number of people polled). It is not fraud, as the tweet implied. It is just a constraint that every field survey operates under, and one that compounds when you are making 294 individual calls rather than one aggregate projection. The results came out right. But that tension between the sample size and the methodology’s ambition doesn’t fully dissolve just because the headline number landed.

Doing it for 4-5 months

There is an important assumption baked into this. If you are collecting opinions over 4-5 months, you are assuming the person’s opinion won’t materially change in that window. That is a risky assumption on paper. But I think it holds in practice more often than not. People don’t really switch their political party of choice in a few months, unless something fairly dramatic happens.

Worth noting though: the SIR controversy (the removal of around 90 lakh voters from the rolls) peaked during the campaign window, well after data collection would have begun. That is exactly the kind of late-breaking structural event that could move opinions, particularly among voters who found their names missing. Whether that showed up in the data or was simply absorbed as noise, we don’t know.

The booth-level sampling within each constituency is another assumption that is hard to audit from the outside. If their 1,500 odd male respondents per constituency skewed toward urban booths, or particular communities, the aggregate swing detection could still be right while individual seat calls were getting lucky. A correct answer arrived at through an imperfect route is still a correct answer, but it is worth knowing which one you have.

For ambiguous problems, you need to absorb some ambiguity by making operational assumptions. That is not a flaw in methodology. It is just how applied work gets done. The alternative is paralysis.

One more thing worth sitting with

There is another data point about this election that I find genuinely interesting, separate from the methodology question.

Before the SIR, West Bengal had more women voters than men: 38.7 million versus 38.4 million. After SIR deletions, that flipped. 5.7 million women were removed from the rolls, against 3.4 million men. Women accounted for 63% of all deletions.

Now, this says nothing about preference. The women who remained may well have voted TMC at exactly the same rates as always. The question of whether 8 in 10 women voted TMC is completely separate from whether there were 38 million or 33 million women on the rolls.

But it does mean that TMC’s female voter base was structurally smaller in 2026 than in any previous election, not because preference shifted, but because a significant chunk of that base was simply not eligible to vote. Whether that was enough to swing seats is something only post-poll data will tell us. The CSDS-Lokniti survey will eventually have the gender crosstabs. Until then, we are speculating.

PPA made their assumptions, committed to them, and got the headline number right. The tweet trying to expose them as fraudulent ended up being a pretty decent advertisement for their approach.

Every model is wrong. This one happened to be useful.

Observations, data, and the not-so-ordinary world. Subscribe if that's your kind of thing.

Subscribe now

Share this post, or tell someone about Ordinary Analysis.

Share

If this series has been useful, subscribe below.

Subscribe now

Ask anyone from West Bengal about the state’s “intellectual heritage”, and they will not stop talking. Tagore. Ray. Sen. Banerjee. Two Nobel laureates in economics. One in literature, also the first Asian. A filmmaker who put Bengal on the world map. A city that ran the country for 138 years.

The data has a more complicated view.

The literacy number

West Bengal’s literacy rate in 2011 was 76.26%. The national average that year was 74.04%. Bengal was ahead and ranked 13th out of 27 states.

Not first or second. Or third. But Thirteenth.

Between 2001 and 2011, that rank did not move. India improved. West Bengal improved at roughly the same pace. No ground gained, no ground lost.1

The gender picture

This is the part of the data that is genuinely positive.

In 2011, the gap between male and female literacy in West Bengal was 11 percentage points. Among 27 major states, West Bengal was placed 9th best on gender parity. Better than Gujarat, Karnataka, Odisha, and most of the Hindi belt. The states with worse gaps included several that are conventionally considered more developed.

By PLFS 2023-24, that gap had narrowed to 6.5 points, placing West Bengal 7th best among 28 major states. The improvement is real, and it is faster than most peers.

On this one measure, West Bengal is in the better half of the country and moving in the right direction. One caveat worth keeping in mind: the improvement reflects a narrowing gap, not both genders becoming more literate. Women are closing in on men. That is a different and perhaps a more meaningful signal than the overall rise in literacy.

Where children fall off

West Bengal and Tamil Nadu are the only two states that maintain their own education MIS and submit bulk data to UDISE+2, rather than feeding directly into the national system. Their primary-level dropout figures, both reported as zero, should be read with that in mind.

The secondary level is a different story.

West Bengal’s secondary school dropout rate in 2023-24 was 17.98%. The national average is 14.1%. Among states above that average, West Bengal sits in the upper half, in the same band as Madhya Pradesh and Chhattisgarh (and surprisingly, Karnataka).3

The structural reason is visible in the school composition data. 79% of schools in West Bengal are at the foundational and preparatory level. Only 11.6% are secondary. Imagine a child who just completed Class 8 in a remote district, and the nearest secondary school is far, then the decision to continue doesn’t remain purely academic. It is logistical and economic. And that often means forcing an undue goodbye.

This is made more uncomfortable by the ghost school data. UDISE+ 2023-24 identified 3,254 schools in West Bengal with zero student enrolment, the highest count in the country. These schools collectively employed 14,627 teachers. Infrastructure and teaching capacity are concentrated where students are no longer located, while districts with growing adolescent populations lack access to secondary education. The state education minister’s response when this was reported: “I have to look into it.”4

Income and poverty

West Bengal has hovered around ₹75-82 for every ₹100 earned by the average Indian over the past decade, consistently below the national average. The gap has not closed.5

We need to take a pause and absorb what this chart is actually telling us. It does not measure absolute income. It measures how each state’s economy performs relative to the national average, on a per-person basis. If the average Indian earned ₹100 in 2023, West Bengal’s economy generated ₹82 per person, Karnataka’s ₹175, Bihar’s ₹36.

That framing matters. West Bengal is not getting poorer in absolute terms. It is failing to keep pace. And in a country where Karnataka and Tamil Nadu are pulling away, failing to keep pace is its own kind of falling behind.

This ties it back to part 2 of the series. When you have an ‘unproductive’ workforce. This is bound to happen.

The multidimensional poverty numbers tell a relatively encouraging story. West Bengal’s MPI poverty rate fell from 21.28% in 2015-16 to 11.89% in 2019-21, below the national average of 14.96%. Welfare delivery has worked. People have moved out of deprivation on the indicators that matter: sanitation, cooking fuel, electricity, and bank access. These are essential for a human being.

But almost every state improved by a comparable margin. West Bengal’s relative position did not change. The same states that were ahead in 2015-16 were ahead in 2019-21. Those who were behind are still pretty much behind. Welfare worked. It just worked everywhere. For everyone.

What the HDI says

The Human Development Index (HDI) tries to answer a simple question: how well are people actually living? It combines various factors that are important for a good life into a single number. India’s state-level HDI rankings place West Bengal at 0.674, ranked 20th among 28 major states, below the national average of 0.685.

West Bengal scores above the national average on both the health component (life expectancy ~71-72 years) and the education component (mean years of schooling ~7.5). Which is not bad. But it is the per capita income that pulls the composite down (might be the same reason why Goa is number 1). Per capita GSDP sits well below the national average.

Amartya Sen and Abhijit Banerjee. The two economists who got their Nobel for measuring and understanding human development are both from West Bengal.6 The state that produced them has a human development index score that Kerala crossed decades ago. That sentence is uncomfortable to write. But it is accurate.

West Bengal delivers at the baseline. Literacy is above average. The gender gap is narrowing. Primary enrolment is high. Poverty is falling faster than in most states. These are real, and the data is quite clear on them.

But the next layer does not hold. Secondary dropout is among the worst in the country. Per capita income is not moving upwards against the national average. HDI has been stuck in the medium band for a long time. The state gets people through the door. It has not figured out what to do with them once they are inside.

Welfare worked. Wealth did not follow.

Data sources:

• Census of India 2001, 2011 (Office of the Registrar General)

• PLFS 2023-24 (MoSPI)

• UDISE+ 2023-24 (Ministry of Education)

• Rajya Sabha Q.1131, Dec 2025 (Ministry of Education)

• NITI Aayog National MPI Progress Review 2023

• RBI Handbook of Statistics on Indian States 2024-25

• Global Data Lab Subnational HDI 2023

If this series has been useful, subscribe below.

Subscribe now

India is the world’s fourth (or fifth) largest economy by nominal GDP. By per capita income, it ranks 136th. That gap is the context for everything that follows here.

At that income level, the ability of most citizens to purchase private healthcare is alarmingly limited. The quality of the public system is not a policy preference. It is essential for survival & wellbeing, which means that when you want to understand how a state is doing, the health numbers are among the most honest data available. They reflect what the system actually delivers to people who cannot opt out of it.

This piece looks at West Bengal’s health profile across five indicators: infant mortality, maternal mortality, anaemia, institutional delivery coverage, and public health financing. Some numbers are good. Others are not. Together, they describe a state that has solved one set of problems and left another untouched.

The IMR picture

West Bengal’s infant mortality rate is 17 deaths per 1,000 live births. That places it alongside Karnataka, well ahead of states like MP, UP, Gujarat, and comfortably below the national average of 25. The improvement over the past decade has been steady and real.

Infant mortality rate measures deaths in the first year of life per 1,000 live births. It is one of the most widely used proxies for a state’s health system because it captures nutrition, maternal care, immunisation coverage, and access to neonatal facilities in a single number. A state that does well on IMR has typically gotten several things right simultaneously. That is quite important for an Indian state.

Across nearly a decade, West Bengal has consistently outperformed India on this measure. The gap has been consistent.

The improvement from 31 in 2013 to 17 in 2023 tracks better than the national pace. On under-five mortality1, West Bengal has already met the 2030 SDG target of under 25. Bihar and Uttar Pradesh are still in the 30s.

The MMR reversal

Maternal mortality tells a different story. A sharply different one.

In 2014-16, West Bengal’s MMR was 101 deaths per 1,00,000 live births. India’s was 130. West Bengal was not just better than average. It was well ahead.

Then something shifted.

India kept improving. West Bengal did not. By 2017-19, the lines had crossed. India’s MMR fell to 103. West Bengal’s had climbed to 109. By the latest available data, India sits at 88. West Bengal is at 104, grouped with Bihar, well above states like Tamil Nadu (54), Maharashtra (33), and Kerala (19).

West Bengal is one of only four states where MMR increased against the national declining trend between survey periods, alongside Haryana, Uttarakhand, and Chhattisgarh2. A peer-reviewed study in Scientific Reports found that while most major states showed a broadly linear decline in MMR from 2014 to 2020, West Bengal showed a cyclical pattern, stagnating where others improved.

The infant survives. The mother does not, at rates that have stagnated over the last few years.

The access explanation does not hold

Poor MMR often stems from poor access to better institutional care. Women delivering at home, without skilled attendance, without emergency obstetric backup. It follows the basic logic.

That explanation does not fit West Bengal’s numbers3.

Institutional deliveries rose from 75.2% in NFHS-4 to 91.7% in NFHS-5. Nine in ten births now happen in a medical facility. Crucially, the cost of delivery in public facilities actually fell over this period, by Rs 5,236 per delivery on average. The access problem, to the extent it existed, has been substantially addressed.

West Bengal got more women into hospitals. But the MMR didn’t improve.

This points to a quality problem, not an access problem. Women are reaching facilities. What happens inside those facilities, and the condition women arrive in, is the unresolved question.

The anaemia evidence

The condition women arrive in is measurable and may be a more important causal factor of why the MMR numbers have stagnated. Anaemia prevalence is one of the strongest predictors of maternal mortality risk. It reduces the body’s capacity to manage blood loss during delivery, increases the risk of preterm birth, and amplifies every other complication.

West Bengal’s anaemia numbers are not just high. They are worsening, and they have been worsening across three survey rounds.

Among women aged 15-49, prevalence increased from 63.2% in NFHS-3 (2005-06) to 62.5% in NFHS-4 (2015-16), and then jumped sharply to 71% in NFHS-5 (2019-21). India’s national average is 57%. West Bengal has the highest prevalence of anaemia among women of any large state in the country.

This is not a story of a state that made progress and then slipped. Anaemia among women was high in 2005, remained high through 2015, and then deteriorated further. The NFHS-4 to NFHS-5 spike of nine percentage points is one of the largest increases recorded among major states in the period.

The men’s numbers add context. Anaemia among men aged 15-49 dropped marginally from NFHS-3 to NFHS-4, from 32.3% to 30.3%, then rose sharply to 39% in NFHS-5. India’s national average is 25%. This is not solely a women’s nutrition story. It is a systemic public health condition that affects the state’s population broadly.

What the budget says

A state with anaemia at 71% among women is a state with a chronic nutritional deficit that predates healthcare delivery. Reversing it requires sustained investment in public health infrastructure, nutrition programmes, and preventive care.

West Bengal allocated 6.4% of its total expenditure to health in 2025-26, marginally above the national state average of 6.2%. On that metric, it appears to be doing its share.

The more telling number is out-of-pocket expenditure. Per capita OOPE in West Bengal is Rs 4,010 annually4, the second highest in the country after Kerala. Over 50% of total health expenditure in the state comes directly from household spending, not government funding.

Kerala’s high OOPE reflects a wealthy population choosing premium private healthcare on top of a functional public system. West Bengal’s high OOPE reflects a public system that is not absorbing enough of the healthcare burden. Patients are paying because the alternative is inadequate.

The 2026-27 interim budget, tabled ahead of the state elections, allocated 7.5% of GSDP to social services spending. Capital outlay, the spending that builds infrastructure, has consistently undershot budget estimates by 20-34% in recent years. What is budgeted and what is built are different numbers.

The summary

West Bengal is a state that has successfully reduced infant mortality to near-southern-state levels. That is a genuine achievement, and the data is clear on it.

The same data shows a maternal mortality rate that stopped improving a decade ago and has since worsened relative to the national average. It shows anaemia among women at 71%, the highest of any large state, rising across three survey rounds. And it shows a public health financing structure where households bear more than half the cost of care.

West Bengal did the work on access. Institutional deliveries are near 92%. The gap is not in whether women reach hospitals. It is in the condition they arrive in, and in whether the infrastructure they reach is equipped to manage it.

The next piece in this series looks at literacy, education, and human development.

Observations, data, and the not-so-ordinary world. Subscribe if that's your kind of thing.

Subscribe now

Share it with someone who you think would enjoy it.

Share

Data Source:

• SRS Special Bulletin on MMR (RGI),

• SRS Statistical Reports

• NFHS-3/4/5 State Factsheets (IIPS/DHS Program)

• PRS Legislative Research Budget Analysis 2024-25 and 2025-26

• National Health Accounts 2021-22 (MoHFW)

• Contributing factors for reduction in maternal mortality ratio in India: https://www.nature.com/articles/s41598-024-65009-0

By nominal GDP, the city clocks in at roughly $150 billion, sitting comfortably behind Mumbai ($310B) and Delhi ($294B) and ahead of Bengaluru ($110B), Hyderabad ($75B), and Pune ($69B). Third place. Top three. Eastern India's undisputed anchor.

The problem is that nominal GDP rank rewards size. Kolkata has 15 million people. It has the eastern port, the rail hub, significant private sector presence in trade, finance, and services. The $150B is not fiction. But the question this series is asking is different: is that number growing fast enough, relative to the city's scale and potential? Is Kolkata punching at its weight, or below it? That’s what we will try to understand here.

This is Part 1 of a 5-part West Bengal Data Series. Subscribe to get the rest directly in your inbox.

Subscribe

Adjust for Population. The Rank Collapses.

Kolkata's metropolitan area has roughly 15 million people. Compare that to Bengaluru (13 million) and Hyderabad (10 million). At those population levels, Bengaluru's $110B GDP yields a higher per-capita output than Kolkata's $150B. Hyderabad, at roughly two-thirds of Kolkata's population, is closing the aggregate gap faster than the headline numbers suggest.

I find the whole numbers game comparable to how India’s GDP is discussed. India, too, has the world’s 5th/6th (ranking is debatable now) largest economy. But just when you look at the per capita numbers, they change drastically. From 5th to 136th. The sum looks powerful; the distribution tells a different story. Kolkata's third-place rank works the same way.

The cleaner picture comes from looking at state-level per capita income, where official data is actually reliable. Delhi's per capita NSDP stands at ₹4.93 lakh. Telangana (Hyderabad's state): ₹3.87 lakh. Karnataka (Bengaluru): ₹3.80 lakh. Maharashtra¹ (Mumbai + Pune): ₹3.09 lakh.

West Bengal: approximately ₹1.7 lakh.

That is not a rounding error. West Bengal's per capita income is less than half of Karnataka's and less than a third of Delhi's. The third-richest city belongs to a state where per capita income sits well below the national average.

Which brings us to another deeper number. Kolkata is the only major Indian city whose share of the national GDP is smaller than its share of the national population.

CPR Insights

This is something worth imbibing. Mega (for India - we now use the term Mega instead of Metro) cities are the places where people move for “better” opportunities. That is to increase their earning potential. So ideally, for a mega city, you should have a population that is more productive. This is clearly not the case for Kolkata. Every other large Indian city — Mumbai, Delhi, Bengaluru, Hyderabad, Pune, Chennai — produces more than its population weight suggests it should.

And yet, Kolkata² is still carrying a slow-growing state.

The New Economy Test

The other way to measure a city's economic vitality is to look at where new wealth is being created: startups, venture funding, and knowledge economy jobs.

In 2024, Bengaluru saw 485 funded startup deals. Delhi NCR, 332. Mumbai, 231. Hyderabad, 56. Pune, 47. Chennai, 46.

Kolkata is not in that tier. According to Inc42's 2024 funding data, Kolkata is categorized alongside Ahmedabad, Jaipur, and Indore as cities with fewer than 100 total funded deals since 2014. It shows up in descriptions as a "regional innovation hub," which is the kind of phrase that sounds encouraging and means relatively little.

The Startup Blink 2024 Global Index places Bengaluru among the world's top 20 cities by startup ecosystem strength, with Hyderabad and Pune inside the top 100. Kolkata does not appear in the top 100.

Office leasing tells the same story more gently. Kolkata posted a 69% year-on-year growth in office space leased in 2025. The highest growth rate in India. But it absorbed 2.3 million square feet in a year that saw Bengaluru absorb 28 million. (For every 1 sq ft of office space that Kolkata absorbed in 2025, Bangalore absorbed 12) The growth rate is real. And a 69% yoy growth is worth acknowledging. But the base is small.

So: What Is Kolkata?

Kolkata remains the commercial and financial centre of eastern India. It is the eastern rail hub, the gateway to the northeast, the port that handles Bangladesh trade. It is not a hollowed-out city. It continues to have real economic activities.

But it is also a city that was built for a different India and has not fully adapted to the one that exists. Its rank at third in nominal GDP reflects both legacy and genuine activity, but the cities below it are growing faster in the things that compound. Some numbers are encouraging in this aspect, but they are far from being assuring. Kolkata needs more of that, and fast.

The state behind the city is the deeper issue. That is what the rest of this series covers: what decisions about industry, investment, and policy produced the numbers above, and what those numbers mean for the 100 million people who live in the state of West Bengal.

This is Part 1 of the West Bengal Election Data Series. Part 2 covers industrialisation, investment, and the long story of Bengal’s economic structure.

If this was useful, subscribe for more on data, cities, and the ordinary world.

Subscribe

Share it with someone you think will like it.

Share the newsletter

Data sources:

• EAC-PM Working Paper WP/31/2024 — Relative Economic Performance of Indian States: 1960-61 to 2023-24 — link

• RBI — Handbook of Statistics on Indian States — link

• NITI Aayog — Macro and Fiscal Landscape of West Bengal (2025) — link

• Knight Frank India — India Real Estate: Office and Residential Market H2 2025 — link

• CPR India — Crudely Estimating Domestic Product for City-Regions — link

• Inc42 — Indian Tech Startup Funding Report 2024 — link

• StatisticsTimes — GDP Growth of Indian States — link

1 This deserves a separate deep-dive. Maharashtra, even after having two mega cities in Pune & Mumbai, is ranked 10th in the per capita income list for states.

2 What we call Kolkata here may differ from study to study. But directionally, it would yield similar results. Some studies would include the industrial belts of Howrah & Hoogly - while others may not.

This is the second piece in Grain of Data's West Bengal Election Data Series, looking at investment and industrialisation. Six comparison states: Gujarat, Maharashtra, Tamil Nadu, Karnataka, Telangana, and Odisha. Different models, different trajectories. WB's position among them is the story.

Subscribe

Where capital goes

Foreign direct investment (FDI) measures the equity capital flowing into a state from overseas. Factories, technological partnerships, long-term investments, and acquisitions. It is not a perfect measure of economic health, but it is one of the cleaner signals of where large investors expect returns.

FDI is an imperfect proxy for investor confidence: it skews toward states with financial hubs (Maharashtra), IT ecosystems (Karnataka). But across five years, it tells a consistent story.

Maharashtra and Karnataka together absorbed the overwhelming share of India's state-level FDI between FY21 and FY25. The five remaining states split what's left, and WB's slice (hovering between $73M and $298M annually) is thin even within that group.

A few things to note before reading this chart cleanly.

Karnataka's FY22 spike to $22B reflects a surge in startup funding rounds (Byju's, Swiggy, Flipkart raises) classified as FDI equity. That capital was real and consequential for Karnataka's tech ecosystem. But it reads differently from the long-horizon manufacturing investment that drove Gujarat's FY24 numbers. Karnataka's FDI has moderated since, reaching $6.6B in FY25 as that funding cycle cooled.

Gujarat's trajectory is the more interesting one. From $1.9B in FY21 to $7.3B in FY24, driven by Micron's semiconductor plant and PLI-linked manufacturing investments. This is long-horizon capital, not startup equity rounds that can exit overnight.

Telangana's FY24 jump of $3B, up 132% year-on-year, was driven by Amazon and pharma manufacturing. A state that inherited Hyderabad's infrastructure in 2014 and has since built on it steadily.

West Bengal's FY25 figure of $298M is its highest in five years. That is a positive signal and deserves fair weight. In absolute terms, it remains at roughly 1.5% of Maharashtra's annual FDI inflow, and 4.5% of Karnataka’s. Almost insignificant in the chart.

Stuck between two models

Not all states are chasing the same economic destination, and FDI alone doesn't capture the structural picture. The chart below plots each state's agriculture share of GSVA¹ (Gross State Value Added) against its industry share, using FY2021-22 data from NITI Aayog's Macro and Fiscal Landscape state briefs.

Gujarat and Odisha sit in the top-left & the top-right: high industry, moderate-to-high agriculture. Both have grown their industrial base significantly. Gujarat at 49.3% industry GSVA is an outlier even among large states nationally. Odisha has taken a resource-and-steel path to a similar position.

Karnataka and Maharashtra sit in the bottom-left: low agriculture, low-to-moderate industry. Tamil Nadu is nearby.

Telangana sits closer to West Bengal than to any other state on this chart. Both carry higher agriculture shares and lower industry shares than their peers. They share a quadrant.

Both sit in the high-agriculture, low-industry quadrant. But Telangana's agriculture share has been declining as its economy grows. West Bengal's has not moved as decisively. High agriculture at 21.1%, comparable to Odisha. Low industry at 24%, comparable to Karnataka. The structural transformation, in both possible directions, is incomplete.

This is not a value judgment. These are the coordinates. What they mean is that WB's economy is organised around a different model, or more precisely, has not yet committed fully to any model.

Manufacturing's signal

Here is where the WB story gets counterintuitive. And it adds to the previous point.

18.8% of West Bengal's workers are employed in manufacturing. That is higher than Karnataka (10.4%), Maharashtra (12.4%), and Tamil Nadu (16.8%). On workforce share alone, WB looks like a manufacturing economy.

The scatter plots manufacturing workers as a share of workforce against manufacturing's share of GSVA. States above the 45-degree line get more output per worker than their employment share suggests: high-productivity manufacturing. States below get less.

West Bengal is below the line. 18.8% of workers in manufacturing, 13.3% of GSVA from manufacturing.

Gujarat sits well above. Gujarat's manufacturing workforce generates 37% of its GSVA from 23.8% of its workers. West Bengal's generates 13.3% of GSVA from 18.8% of workers. The gap in manufacturing productivity is visible even from these ratios.

The gap is explained by composition. 47.12% of West Bengal's informal enterprises are in manufacturing, the highest share in India (ASUSE 2023-24). West Bengal alone employs 40% of all women working in tobacco manufacturing nationally. Half of India's unorganised tobacco enterprises are in the state. The average registered MSME in WB employs 10 workers. For Telangana, that figure is 143.

WB's manufacturing is wide but shallow. Bidi, handloom, jute, cottage metalworks. These are industries with high labour intensity and very low value addition per worker. They show up in the employment data without showing up proportionally in output.

The manufacturing growth rate of 8.1% per annum over the last decade is real, higher than the national average of 5.5%. But it is occurring primarily in a thin formal layer on top of a vast informal base that has not been upgraded.

The formal economy that never scaled

The IT exports chart is the starkest in this series.

Karnataka exports ₹6.20 lakh crore in software annually. West Bengal exports ₹0.36 lakh crore. The ratio is roughly 17:1, with a state accounting for 2/3 of West Bengal’s population.

Kolkata had, and still has, credible inputs for an IT economy (as we can see - more than Gujarat & Odisha). Salt Lake Sector V and New Town have established IT parks. Real estate costs remain lower than Bengaluru or Hyderabad. The English-speaking graduate workforce is large. TCS, Wipro, and Infosys all have established presences.

The formal economy of WB (the organised manufacturing sector, the IT sector, the parts that generate corporate tax, attract large-scale private investment, and create salaried employment) has not grown fast enough to become the dominant story. West Bengal’s economy is disproportionately concentrated in sectors that are difficult to tax and difficult to attract large-scale private capital into: informal manufacturing, low-productivity services, and agriculture.

What the data says

The four charts tell one consistent story - if we examine closely.

Capital flows to Maharashtra and Karnataka at a scale West Bengal has not approached in any recent year. The structural position shows an economy carrying high agriculture and low industry, having not moved decisively toward either an industrial or a knowledge-services model. Manufacturing employment is large but productivity per worker is low, concentrated in informal micro-enterprises rather than factory-scale production. And in the formal knowledge economy, the gap with the leading states is not narrow.

None of this makes West Bengal a failing economy (at least, yet). Its manufacturing is growing faster than the national average. Its FDI is at a five-year high. It absorbed ₹4,416 crore in central capital investment loans, the fifth highest among all states. The public sector capacity and willingness to invest is present.

What is less present is the private sector conviction that follows. The investment data, the structural position, the manufacturing productivity gap, and the IT export numbers all point in the same direction. West Bengal is a large economy organised around sectors that are hard to scale, hard to tax, and hard to attract large-scale private capital into.

That is where things stand today. Whether the next decade changes the composition is an open question.

The next piece in this series looks at health parameters.

If this was useful, subscribe for more on data, cities, and the ordinary world.

Subscribe

Share it with someone you think will like it.

Share Ordinary Analysis

Data Sources:

• Sectoral GSVA shares from NITI Aayog Macro and Fiscal Landscape state briefs (2025), based on MoSPI data for FY2021-22.

• FDI figures from DPIIT quarterly factsheets; some FY21-23 figures are approximate.

• Manufacturing workers from PLFS 2022-23 via NITI Aayog briefs.

• IT exports from STPI and SEZ official data, FY2023-24.

• MSME firm size from DC-MSME Udyam Registration Bulletin. Informal enterprise share from ASUSE 2023-24 via Factly.

1 GSVA (Gross State Value Added) is the total value of goods and services produced within a state, after subtracting input costs. Think of it as the state's contribution to national output. If a factory in Gujarat buys ₹60 worth of raw materials and sells output worth ₹100, it adds ₹40 to Gujarat's GSVA. It is the primary measure used to calculate GSDP.