India's Bond Yields Are Rising. The Trade Data Says It's Not the Growth Story You Think.

Last week The Economist posed a two-theory question about rising global bond yields: are markets pricing in fiscal profligacy and inflation, or an AI-driven capital-expenditure cycle that pits private borrowers against governments for the same pool of savings? It offered India as a test case. India's 10-year G-Sec yield rose 67 basis points this year to roughly 7% — comparable to the US Treasury selloff. GDP grew 7.8% year-on-year in the June quarter. Private investment reportedly jumped 12%. New project announcements, particularly data centres and nuclear plants, were up 27%.

The article ran the tests. On fiscal discipline, India passed: inflation averaged 4.6% since the RBI adopted its target, versus 8.1% the prior decade. The primary deficit is back below 2% of GDP. On the growth theory, the article called it a wash — because India's bond market is captive. Domestic insurers, banks, and pension funds are mandatory buyers. Foreign ownership of G-Secs sits at roughly 3%. Bond yields cannot price growth expectations if the market itself cannot price anything.

The article concluded: "institutions matter."

The argument is tidy in the way financial-economics arguments often are: it tests a financial theory using only financial data, finds the financial data compromised, and closes the case. But trade data offers a different kind of test — one that decomposes the economy rather than the bond market. And that test does not return "inconclusive." It returns two answers at once.

The Merchandise Half: Theory A

Start with what the goods-trade data shows. India's manufacturing value-added as a share of GDP was 17.9% in 1995. It was 17.0% in 2010. In 2024, it was 13.1% — the lowest figure since at least 1960. This is not a country in the grip of a capital-investment boom, at least not of the kind that builds factories.

The merchandise trade deficit tells the same story with larger numbers. In 1995, India's goods deficit was $5.9 billion. In FY2025, it hit $284 billion — a 48-fold increase over a period when nominal GDP grew 10-fold. The deficit is expanding nearly five times faster than the economy.

Foreign direct investment, which should surge during a capital-expenditure cycle, has done the opposite. FDI peaked at $66 billion in 2022, driven by Google, Meta, and Apple's early supply-chain commitments. By 2024, it had collapsed to $27 billion — a 59% decline in two years. China, for all its well-documented FDI troubles, still drew $43 billion.

The China dependency is deepening, not narrowing. India imported $127 billion from China in 2024. China's share of Indian imports rose from roughly 2% in 1995 to 19% in 2024 — despite tariffs, anti-dumping duties, quality-control orders, and five years of Production-Linked Incentive schemes explicitly designed to reduce it. India's bilateral deficit with China alone reached $112 billion, representing 35% of India's entire trade deficit.

If a bond market existed in a country where manufacturing share had fallen for three decades, where the trade deficit expanded five times faster than GDP, and where the largest bilateral trade relationship was a one-way dependency on a geopolitical rival — that bond market would be pricing fiscal and external risk. This is the half of India's economy that supports Theory A.

The Services Half: Theory B

Now decompose the data differently. India's services value-added crossed 50% of GDP in 2019 and sits at roughly 49% today. This is not a new phenomenon — services have accounted for more than 40% of Indian GDP since 1998 — but the composition has changed. IT services exports reached $246 billion in FY2026, up from $136 billion in FY2019, an 81% increase in seven years, according to NASSCOM.

The services trade surplus was $189 billion in FY2025, offsetting two-thirds of the $284 billion merchandise deficit. The combined goods-and-services deficit narrowed to $94 billion. India is not, in other words, simply borrowing to consume. It is importing goods — electronics, machinery, components — and exporting services at a scale that substantially offsets the bill.

The data-centre buildout provides the physical counterpart to the services-export story. Google announced a $15 billion investment in a 1-gigawatt data centre in Andhra Pradesh in late 2025. Amazon Web Services committed $12.7 billion by 2030. Microsoft pledged $3 billion. TCS, India's largest IT services firm, announced a $6.5 billion data-centre plan. Cumulative ecosystem investment is projected at $10–12 billion annually through 2030. India's live data-centre capacity, roughly 1.35 gigawatts today, could triple by the end of the decade.

This is the mechanism the user's critique pointed to, and the data supports it. IT services firms, telecom operators, and hyperscalers are borrowing to build digital infrastructure. They import servers, GPUs, and networking equipment — much of it from China and Taiwan, feeding the merchandise-import surge. They hire and train engineers. They export cloud services, AI capabilities, and business-process outsourcing to the West. The investment is real. It just never touches a factory floor.

Supporting this read: India's resident patent applications have grown from roughly 20,000 in 2010 to 85,000 in 2024 — the fastest growth rate among major economies. Container port throughput nearly tripled from 9.2 million TEU in 2010 to 23.9 million TEU in 2024. Goods are moving. Innovation output is rising.

One data point resists the growth narrative. India's R&D spending as a share of GDP has declined from 0.86% in 2008 to 0.65% in 2020. An AI investment boom that does not raise economy-wide R&D intensity is a boom in deploying existing technology, not in creating new one. This is a distinction worth watching — but it does not invalidate the investment story. It qualifies it.

The Split Verdict

The Economist framed India as a test case between two theories and called the test inconclusive because the bond market is captive. The trade data says the test is not inconclusive. It is split.

India's economy is running two trajectories simultaneously. On the merchandise side: a persistent under-exporter, dependent on Chinese industrial inputs, running a structural deficit that looks like a fiscal-risk signal. On the services side: a genuine digital-infrastructure investment cycle, producing export revenues that substantially offset the goods deficit, looking like the growth story the Bessent theory describes.

The bond market, if it were open, would not produce a clean verdict. It would price both signals at once — fiscal risk from the import surge, growth expectations from the services-export trajectory — and the net effect on yields would be ambiguous. The Economist's framing assumed the bond market would pick one theory or the other. The data suggests a bond market would be confused for good reason.

What to Watch

Three indicators will determine whether the services-investment arc eventually pulls the merchandise arc with it — or whether the two halves of the Indian economy continue to diverge.

First, the services surplus trajectory. The $189 billion surplus in FY2025 offset 67% of the goods deficit. If that ratio continues to rise — toward 80%, then 90% — India's external accounts begin to look like a country funding its own investment rather than borrowing for consumption.

Second, the import composition. Electronics imports from China are $47.7 billion and machinery $27 billion. If an increasing share of those imports can be shown to feed the data-centre and telecom buildout — rather than consumer electronics and smartphones — the "investment boom" story gains specificity. The data to make this call exists but is not publicly disaggregated at the necessary level.

Third, the R&D number. A digital-infrastructure boom that does not eventually raise economy-wide R&D intensity above 1% of GDP is an adoption story, not a capabilities story. The difference matters for whether India's services exports are defensible or vulnerable to the next shift in the global technology stack.

For now, India's bond yields are rising for reasons that both theories capture — but neither captures alone.

---

Data sources: OECD Main Economic Indicators via FRED (trade), World Bank WDI (GDP, manufacturing share, FDI, R&D, services share), India Ministry of Commerce & PIB (bilateral China trade, services trade surplus), NASSCOM (IT services exports), US Census Bureau (bilateral US-India goods trade), WIPO (patents), DGCI&S (combined goods+services deficit), OECD TiVA (value-added shares).