Rethinking India’s Growth Story

ELI5/TLDR

India’s official GDP numbers may be wrong in both directions. The new paper by Abhishek Anand, Arvind Subramanian and Josh Felman argues growth was underestimated by about 1.5 points in 2005-11 and overestimated by 1.5-2 points in 2012-23. If they’re right, the recent “fastest-growing major economy” run was really closer to 4-4.5% a year, not 6%. The episode then fans out into three related complaints: the RBI propping up the rupee, electricity priced so badly it taxes factories at ~85%, and free-trade deals that get quietly cancelled out by anti-dumping duties.

The Full Story

The smell tests that started it

Anand doesn’t lead with the math. He leads with two things that just don’t add up.

First, the macro indicators. Compare 2005-11 to 2011-23 and almost everything important falls off a cliff — exports, imports, investment, bank credit, FDI.

All of these indicators, key macro indicators, they have collapsed by a fairly large magnitude… you would expect that if there is a collapse in each and every macroeconomic indicator, the growth will fall as well. But you don’t see that in data.

Second, demonetisation. In one night 85% of the currency left circulation — the largest monetary shock an emerging market has taken in recent memory. You’d expect nominal GDP to flinch.

Your nominal GDP, it surges from around 9-9.5% the previous year to around 11% the year demonetisation happened.

Currency vanishes, the economy supposedly accelerates. That was enough to make him look harder.

Problem one: using the formal sector as a stand-in for the informal

India can’t directly measure most of its informal economy, so it assumes the informal sector grows at the same rate as its formal counterpart. Formal manufacturing comes from the Annual Survey of Industries; informal manufacturing is just assumed to track it. Informal transport gets proxied off motor-vehicle sales. And so on.

This was a defensible shortcut once. The paper shows that pre-2011, formal and informal growth rates moved together, so borrowing one for the other introduced little error. Then came the triple shock — demonetisation, GST, Covid — each of which hammered the informal sector specifically while leaving the formal sector comparatively intact.

Now the formal sector and the informal sector, the growth rates have diverged completely, which was not the case earlier.

So the proxy keeps assuming the two move in lockstep at exactly the moment they stopped. Health and education are worse: their informal growth is extrapolated from household consumption surveys, and the crucial 2017-18 consumption survey was junked and never released. The next usable one only arrived in 2022-23. For years the deflators were being driven off 2005-11 spending patterns.

Problem two: deflating with the wrong prices

About 75-80% of GDP is computed in nominal terms first, then deflated to real terms using sector price indices. The trouble is which index. India deflates large chunks of services — trade, hotels, restaurants, construction — using the Wholesale Price Index, a basket of industrial inputs like steel, cement and oil.

WPI is nothing but basically it’s tracking oil prices, and that’s the problem.

Using a stale, oil-dominated wholesale basket to convert restaurant revenue into “real” output is a category error. The right tool would be something closer to a consumer or producer price index — the price people actually pay, not what wholesalers pay for steel. When the deflator is wrong, real growth is wrong.

The numbers, and the unusually generous tone toward MOSPI

The break point is 2011-12 because that’s when the new base year kicks in. The verdict: 2005-11 underestimated by ~1.5 points (official ~6.5%, should be ~8%); 2011-23 overestimated by 1.5-2 points (official ~6%, should be ~4-4.5%).

What’s notable is how much credit Anand gives the statistics ministry. The informal-sector survey now runs annually instead of lapsing for years; a construction-sector pilot has started; CPI has been updated; the latest GDP revision was genuinely consultative with outside experts. He frames his own work as complementary — something MOSPI can lean on when it builds the back series. This is not a hit job on the agency; it’s a critique of a methodology the agency is itself already repairing.

The rupee: a hidden, broad-based tax

On exchange rates, the panel is critical of the flat-as-a-board rupee from mid-2023 to 2024, kinder to current governor Sanjay Malhotra for letting it float more. The mechanics of defending a currency: the RBI sells dollars and buys rupees, which drains rupee liquidity, which pushes interest rates up — exactly the wrong move in a country forever wondering why private investment won’t fire.

The deeper argument is elegant. Letting the rupee find its level is a better industrial subsidy than the PLI schemes, because it’s uniform.

If you let the currency depreciate, it makes imported inputs more expensive, and that’s a uniform subsidy to any industry in India… you don’t run into this problem of big groups lobbying.

No picking winners, no rent-seeking. On the iPhone objection — won’t a weaker rupee raise the cost of imported components? — Anand flips it: a cheap rupee gives Apple a reason to push suppliers to localise rather than import, which raises domestic value addition. He’s blunt that the jobs aren’t coming from electronics or services anyway; they’re in apparel, leather, footwear, textiles, food processing — thin-margin sectors where the exchange rate is decisive, and where the China-plus-one window is real but narrow.

Electricity: an 85% tax on factories that mostly subsidises the rich

The power-sector segment lands the hardest. The standard subsidy figure is an accounting fiction. Everyone computes subsidy as the gap between the regulator’s tariff (say 4 rupees) and free supply. Wrong reference point, Anand says — what matters is the cost of production, around 9-10 rupees a unit.

The subsidy that we are giving… is not 4 rupees but 10 rupees.

That hidden cost gets recouped by overcharging industry and commerce, who end up paying an effective tax of roughly 80-85% per unit. For energy-hungry sectors — textiles, data centres — that’s disqualifying, which is why data centres now demand bespoke deals or go off-grid entirely (Google and Reliance have been granted parallel licences in Andhra Pradesh).

Then the kicker on who actually benefits. Free-power schemes (100/200/300 units) have made household subsidies rival agricultural ones — together ~45-50% of total power subsidy. But where a state gives 200 free units, 85-90% of households consume less than that and pay nothing. Since not 90% of India is poor, most of the subsidy flows to the middle class and the rich. In Punjab, wealthy farmers dominate consumption, so maybe 15% of the agricultural subsidy reaches poor households and 85% goes to rich ones. The fixes — privatisation (Odisha), competition, letting big industrial users exit the grid — all attack the same root: distribution companies are unaccountable monopolies with zero incentive to be efficient.

Trade: free on paper, protected in practice

The FTA argument is that India keeps neutering its own deals. The ASEAN-India agreement should have given apparel makers zero-tariff access to Indonesian viscose — a key input as the world shifts from cotton to synthetics. Instead, just before the deal took effect, India slapped anti-dumping duties on Indonesian viscose, “temporary” for six months, extended for years.

On paper you have 0% tariff, but… you still have to pay anti-dumping duties, and that basically meant it nullified whatever the gains were expected to be.

The beneficiary was a single near-monopoly domestic producer, which then charged Indian buyers more than it charged the world. No infant-industry case — just rent extracted from a competitive, job-rich downstream apparel sector. When anti-dumping duties expired, Quality Control Orders (QCOs) replaced them — a blunter instrument that simply blocks imports.

Anand’s defence of FTAs is that they were never allowed to work, so “FTAs are bad” is an unfair verdict. On the recent rollback of QCOs and tariffs, he refuses to call it reform.

These are all self-inflicted pain… in 1991 we learned and we course-corrected. There is nothing new to be learned and gained from again going back to that era.

His worry is durability: much of the current opening is driven by external pressure, especially from the US. Once the global pressure eases, he’s skeptical the protectionist reflex won’t return.

Key Takeaways

The headline claim: Indian GDP under-counted by ~1.5 points in 2005-11, over-counted by 1.5-2 points in 2012-23 — real recent growth closer to 4-4.5% than 6%.
Two mechanical causes: proxying informal-sector growth off the formal sector (broke when the triple shock hit informal harder), and deflating services/construction with an oil-dominated, outdated Wholesale Price Index.
A weaker, market-priced rupee acts as a uniform, non-distorting industrial subsidy — better than PLI, which picks winners and invites lobbying.
Electricity priced below cost for households is recouped as an ~85% effective tax on manufacturing; most of the household subsidy reaches the middle class and rich, not the poor.
FTAs get cancelled out by anti-dumping duties and QCOs — the ASEAN-India viscose case enriched one monopoly producer at the expense of the apparel industry.
MOSPI gets genuine credit — the critique targets the old methodology the agency is already fixing.

Claude’s Take

This is a high-signal hour, and the rare macro conversation where the host actually makes the guest explain the mechanism instead of nodding along. The GDP claim is the headline, but it’s also the most contested — three economists with a known prior (Subramanian has argued overestimation since 2015) producing a result that confirms it. The smell tests are persuasive precisely because they’re not technical: collapsing macro indicators alongside steady GDP, and a currency shock that somehow accelerated nominal growth. Those are hard to wave away. Worth holding lightly until MOSPI’s back series lands, which Anand himself frames as the real test.

The non-GDP segments are where it earns its keep. The electricity argument — that the true subsidy is measured against cost of production, not the regulator’s tariff, and that it’s a regressive transfer dressed as poverty relief — is the kind of reframing that changes how you read every “free power” election promise. The rupee-as-uniform-subsidy point is clean economics, and the answer to the iPhone objection is genuinely clever rather than evasive.

Two caveats. The protectionism critique is fair but one-note; “we already learned this in 1991” is rhetorically strong and analytically thin, since the global trade environment of 2026 is not 1991’s. And the whole thing has a coherent ideological frame — float the rupee, price power at cost, drop the trade barriers — that’s mainstream liberal economics. Reasonable, but it’s a worldview, not a neutral readout. Docked a point for the GDP claim’s contested status and the occasionally transcript-garbled delivery; otherwise this is among the better India macro hours available. An 8.

Rethinking India's Growth Story