Economic impacts of tax reductions

It has become a canon of economic orthodoxy that lowering corporate taxes will spur investment, and lowering income and indirect tax rates will spur consumption and economic activity, and both will boost tax revenues. 

India’s experience with several direct and indirect tax reforms over the last decade may be a test case to evaluate this orthodoxy. This post analyses the impact of these reforms on tax revenues and the economy, especially of corporate taxes. 

On 20 September 2019, the government slashed the base corporate tax rate from 34.94% to 25.17% (inclusive of surcharge and cess) for existing domestic companies, and to 17.16% for new manufacturing companies — representing a fiscal cost of roughly ₹1.45 lakh crore or 0.7% of GDP.

AK Bhattacharya has this description of the corporate tax rate reductions.

In 2016-17, new manufacturing companies incorporated on or after March 2016 were given the option to be taxed at 25 per cent plus surcharge and cess (compared to 30 per cent plus surcharge, etc.) if they did not claim profit-linked or investment-linked deductions, investment allowances, or accelerated depreciation. Additionally, the tax rate for all companies with an annual turnover of less than Rs 5 crore was brought down to 29 per cent plus surcharge and cess. In 2017-18, the tax rate for small and medium companies with an annual turnover of up to Rs 50 crore was brought down to 25 per cent. This meant about 96 per cent of companies that filed a tax return were brought under a concessional tax rate of 25 per cent plus surcharge and cess... In the following year, 2018-19, the government extended the coverage of the 25 per cent tax rate to cover all companies with an annual turnover of up to Rs 250 crore — a move that would benefit 99 per cent of companies filing tax returns... In 2019-20, the government extended the concessional tax rate of 25 per cent to all companies with an annual turnover up to Rs 400 crore, thereby covering 99.3 per cent of all companies filing tax returns. Subsequently, in September 2019, all companies not availing themselves of the various exemptions and incentives like tax holidays were allowed to be taxed at 25 per cent, inclusive of the 10 per cent surcharge and a 4 per cent cess. Moreover, manufacturing companies starting operations after October 1, 2019, were to be taxed at an overall rate of 17 per cent.

And this on its outcomes in terms of revenue impact.

Corporation tax collections used to be about 34 per cent of the Centre’s gross tax revenues in 2014-15. This share plummeted to 28 per cent in 2019-20 and further down to 23 per cent in 2020-21... the share of corporation tax in GDP has kept falling almost every year in this period — from 3.4 per cent of GDP in 2014-15 to 2.74 per cent in 2019-20 and 2.28 per cent in 2020-21... In 2014-15, about 188,000 companies in a sample size of close to 580,000 paid taxes at an effective rate of over 30 per cent and this cohort accounted for 60 per cent of the total corporation tax collected by the Centre that year. In 2018-19, thanks to the various tax concessions, only about 85,000 companies of a larger sample size of 885,000 paid taxes at the rate of over 30 per cent. And this cohort accounted for only 50 per cent of the corporation tax revenue of the Centre... In contrast, there were just about 24,000 companies in 2014-15 paying taxes at an effective rate of 25-30 per cent, accounting for only 16 per cent of the corporation tax collected by the government. Another 15,000 companies paid taxes at the rate of 20-25 per cent, but their contribution to the corporation tax revenue was only 10 per cent. By 2018-19, the number of such companies saw a huge increase, without, however, a corresponding increase in their share in total taxes collected. Companies paying taxes at 25-30 per cent numbered around 184,000 in 2018-19, but their share in corporation tax was 19 per cent. The number of companies paying tax at 20-25 per cent increased to over 46,000 and their share in total corporation tax rose to 23 per cent.

In short, the story of India’s corporation tax revenues is about how more and more companies have been taxed at a lower rate. As a result, the contribution of a large number of companies to the corporation tax kitty is getting smaller. No wonder, corporation tax buoyancy has suffered in the last seven years.

What has been the impact of corporate tax reductions on the wider economy? 

In absolute value trends, there’s no perceptible spike in private sector gross fixed capital formation (GFCF) from reductions in the corporate tax rates. Nor is there any Laffer curve-type rise in corporate tax revenues.

The twin objectives of corporate tax reductions are to spur private investment and, through it, drive up corporate tax collections. However, private sector GFCF as a share of GDP has fallen sharply since the two corporate tax cuts, and corporate tax revenues as a share of GDP have also been declining. It was at 2.98% of GDP in FY25 compared to 4% in FY12!

In fact, even as profits after tax rose as a share of GDP, the share of private capex in private sector GFCF declined and has been on a secular decline since 2010-11, apart from a slight recovery from the Covid-19 dip. This should be a matter of concern. 

After these reductions, manufacturers in India face one of the lowest corporate tax rates globally today, above only a tiny number of countries. The rates are lower than all major developing country peers. 

How do personal income tax (PIT) collections look? Encouragingly, it has been rising impressively since the regime shift in FY21. In fact, PIT collections have gone from ₹1.69 LC in FY11 to ₹11.83 LC in FY25 — growing faster than corporate tax every year since FY20, which is a structural reversal.

This performance on PIT is confirmed by its sharp increase as a share of GDP since the regime shift. Clearly, the PIT reforms (coupled perhaps with better detections and enforcement) have been a resounding success in terms of increasing PIT from 2.51% of GDP to 3.58% of GDP, a spectacular 43% increase as a share of GDP over just four years. 

However, as a share of PFCE, the trend has been muted, remaining range-bound in the 60-62% of GDP, though there’s a slight uptick since FY24, which must be watched for sustainability.

On the indirect taxes front, the revenue collections response to rationalisation and reductions have not been encouraging. 

Bringing all of them together, if we take out the pandemic volatility, the decade of direct and indirect taxation reforms and rate reductions (eight episodes in total) does not appear to support the economic orthodoxy on tax reduction’s impact on investment, output, and revenues. 

The launch of GST in July 2017 initially caused a dip in indirect tax collections in FY18, as businesses adjusted. The subsequent compliance surge (event F, FY22 onward) then drove indirect taxes to record levels — monthly GST collections averaging over ₹1.5 lakh crore from FY23.

The GFCF line tells a sobering story. From a peak of 36.5% of GDP in FY11, it fell steadily to 26.9% in FY20 and has recovered only modestly to about 29.5% in FY25 — still well below the FY11 peak. Neither the corporate tax cut (D) nor the post-COVID rebound has restored investment to its earlier trajectory, a point highlighted here.

The most notable pattern is that corporate taxes have shown the highest volatility among all tax revenues. 

Post-COVID (FY22 onward), PIT has consistently grown 20–40% annually — far outpacing corporate tax — making personal income tax the engine of direct tax buoyancy. GST's 29% surge in FY22 and 22% in FY23 (H region) may be a reflection of the compliance dividend of digitisation and e-invoicing, more than any rate changes.

The positive revenue response of PIT compared to corporate tax is perhaps due to corporate tax cuts being a one-time rate reduction that reduced the base but boosted profitability without proportionally boosting investment, while PIT benefits from a widening formal employment base. It is also perhaps understandable that the problem with corporate tax is less about base expansion and more about avoidance and evasion, neither of which is directly addressed through rate reductions. 

The private corporate sector's savings have consistently gone up — from 9.5% of GDP in FY12 to over 11% — while its investment as a share of GDP has been falling, indicating that companies have been using higher post-tax profits to build reserves rather than to invest in fresh capacity. It is not incorrect to argue that the 2019 corporate tax cut largely transferred fiscal resources to shareholders and balance-sheet strengthening rather than to productive fixed investment. This also says something about the economy’s aggregate demand growth expectations.

This experience on corporate tax reductions is in line with global experience. A meta-study of 441 estimates from 42 primary studies by Sebastian Gechert and Philipp Heimberger corrected for publication bias — which favours reporting growth-enhancing results — and found that the average effect of corporate tax cuts on GDP growth cannot be rejected from zero. The raw literature (before correction) shows 68% of studies finding a positive effect, but once publication selectivity is accounted for, this falls to about 38%, with nearly half finding a neutral result.

Interestingly, they show that “it is about 2.7 to 3 times more likely to publish a result showing a statistically significant positive impact of corporate tax cuts on growth compared to a significant negative result.” This positive bias is a big problem across economics and elsewhere, and does not get the attention it deserves. 

Their main findings are worth quoting:

First, corporate tax cuts tend to be even less growth friendly when considering a short time horizon. Second, considering both rate and base changes by looking at an effective average corporate tax rate may lead to slightly more positive growth rates in response to tax cuts. However, this is an outlier as compared to the rest of the literature using effective marginal tax rates, corporate tax shares in GDP or statutory tax rates, and the result is also not entirely robust to variations in the meta-regression estimator. Third, there does not seem to be a substantial difference between OECD and non-OECD countries regarding the growth effects of corporate tax changes. Fourth, explicitly controlling for other types of taxation (personal income taxes, capital income taxes, property taxes, sale taxes) does not affect our main findings. Fifth, more recent studies tend to find less growth enhancing effects of corporate tax cuts. Finally, it matters what happens to other budgetary components in conjunction with a corporate tax change: if we hold government spending fixed, a corporate tax hike will be slightly more detrimental to growth, implying that using the additional revenues for government spending instead of fiscal consolidation may foster growth, in line with theoretical arguments from endogenous growth models and empirical evidence on substantial productivity of public capital.

The key conditional findings are: corporate tax cuts reliably attract FDI (semi-elasticity ~-2.9, robust across studies), but their effect on domestic investment is weaker. When product markets have imperfections — which is increasingly the case given rising corporate market power — firms respond to tax cuts by increasing savings and reserves rather than investment. India’s equivalent experience — corporate PAT rising to record levels while private GFCF as a share of GDP fell — is a near-perfect illustration of this theoretical mechanism.

See also this earlier blog post on the questionable virtues of lowering corporate tax rates.

To conclude, let me add to Dan Neidle’s nice description of the tax populism of right and left-wing politicians. 

The tax populism of the right is that we can cut tax without anyone (or at least anyone the populists care about) being hit by cuts in services or benefits. There’s a magic money tree of government waste that can be harvested without consequence. The tax populism of the left is that we can fund services without anyone (or at least anyone these populists care about) being hit by increased tax. There’s another magic money tree, where trade-offs don’t exist.

The enduring tax populism among economists (and market experts) is that we can cut taxes and harvest a triple-win of an increase in investments, a rise in output, and higher revenue collections. There are no such free lunches. India is only the latest in the series of data points that invalidate this tax populism.

Thoughts on international development IX

This is in continuation to the posts consolidated here and a subsequent one here. 

In the case of human-engagement intensive and quality-based interventions (or thick activities), the conventional wisdom in development revolving around evidence-based policy/program formulation and their planned implementation generally fails when the rubber hits the road. Instead, they require starting with some basic version of a program (a minimum viable product), then iterating intensely during implementation, and opportunistically building elements on the MVP that increase the likelihood of success. 

Apprenticeship promotion programs are a good example. The Business Standard has an article on India’s apprentice programs.

Apprentice programs have been a constant part of skill development and job creation initiatives of the central and state governments for several decades. In recent times, there have been the National Apprenticeship Promotion Scheme (NAPS) and the National Apprenticeship Training Scheme (NATS). The news article has this assessment from a recent NITI Aayog report on the country’s apprenticeship ecosystem;

India’s apprenticeship ecosystem remains fragmented and uneven. In 2024-25, while 1.31 million candidates registered for apprenticeship, only 985,000 were engaged and barely 251,000 completed their training — exposing significant leakages between enrolment, engagement, and completion… Medium and large enterprises constitute fewer than 30 per cent of active establishments but account for over 70 per cent of apprenticeship engagement… Weak linkages between educational institutions and industry further undermine programme effectiveness… Women account for only 18.2 per cent of the apprentice pool.

It makes these suggestions

The report recommends establishing a consolidated national-apprenticeship mission, which would serve as an umbrella framework for all apprenticeship initiatives. It envisions a single-window digital interface called the national apprenticeship portal, which integrates information on diverse apprenticeship programmes and provides streamlined access through one common platform. It recommends creating an apprenticeship-engagement index to benchmark state performance, standardising evaluation and assessment protocols, and empowering district skill committees as local anchors. To widen participation, it proposes introducing an apprenticeship-linked incentive scheme that provides financial incentives to both employers and apprentices, particularly targeting aspirational districts, the Northeast, and women apprentices.

For industry, it suggests cluster-based consortia of micro, small, and medium enterprises, a startup apprenticeship programme, and expansion into gig and sunrise sectors such as electric mobility, green energy, and digital services. Targeted incentives for aspirational districts and women apprentices, alongside post-training support and social-security coverage, can be critical for improving retention and completion.

It finally suggests that if effectively implemented, these reforms can ensure the successful adoption of apprenticeships in the country. I’m not sure. 

This is a good example of an intervention where inputs and processes built on even the most rigorous evidence base, and meticulous and prescriptive implementation planning cannot ensure success. It is also an illustrative example of how the discrete logistical approaches crowd out more organic engagement.

The NITI Aayog report proposes several elements of logistics - an umbrella framework, a single-window portal, an index to benchmark performance of states, standardisation of evaluation protocols, creation of district skill committees, apprenticeship-linked incentives to both apprentices and employers, targeting of women and backward areas, etc. 

These elements give the form of a successful intervention, without actually ensuring desired outcomes. While such logistics elements are essential ingredients, they are not sufficient in the case of thick interventions that require behaviour and culture changes and where the quality of engagement is critical. 

Worse, the overlay of well-intentioned, tightly prescriptive planning erodes local discretion and flexibility, and straitjackets the implementation. By crowding out the most critical ingredient for their success, local ownership and adaptive refinement, the program struggles. 

So what could be done? For a start, we must acknowledge that such interventions cannot be one-size-fits-all nationwide programs. Second, the standard method of detailed prescriptive guidelines that revolves around implementation logistics cannot ensure effective implementation. Third, we must eschew the obsession with headline numbers and massive scale from the get-go. Such initiatives can at best scale gradually. Finally, this would require a single-minded focus on outcomes that involve changing habits and practices on the supply and demand sides. Apprenticeships should become attractive to both students and employers. 

All these are hard by themselves and more so for paradigms defined around national programs. One approach could be to start in a few places by identifying firms that are already hiring apprentices or have an interest in doing so (how do we elicit this?), and engaging actively with them to declog the supply side of apprentices and ease their program access. It would also be required to work with the supply side, including educational and training institutions, to encourage and make apprenticeships attractive and desirable. This would generate a few illustrative examples of successes that can then act as lighthouses for emulation and scaling. Besides, it would also give insights about what works and does not. 

Further, these are high-human-engagement intensive activities, whose success would depend critically on internal champions. Bureaucratic processes and practices alone will struggle to achieve the desired outcomes. They can only be complements to the main task of local engagement and struggles, driven by internal champions. 

Other such examples of interventions would include those aimed at improving student learning outcomes, improving public health and sanitation, increasing the adoption of vocational education, the adoption of public-private partnerships in social sectors (education, health, community assets, etc.), etc. 

Unlike programs that deliver specific inputs with defined processes and have a clear theory of change, these examples require behavioural and cultural shifts. No amount of top-down logistics-based inputs and processes can be a substitute for the hard grind of the bottom-up systemic engagement and collective struggles required to shift behaviours and cultures. 

The logistics/procedure recipe in all these cases is mostly well-known. But there’s a never-ending search for ideas and innovations that can add to the recipe. Instead, the objective should be to bring them all together in a manner that ensures success. This is generally about implementation with a basic design, continuous iteratation, and opportunistic refinement of the design. 

In this context, it is appropriate to conclude from an earlier post that, like other walks of life, implementation is all that matters.

The fundamental insight is that it’s not ideas that lead to development but their implementation, and that implementation is almost always far more daunting than the process of discovery of the idea itself. In fact, only a fraction of the pipeline of ideas ever finds its way into successful implementation… The most valuable individual and collective attributes for progress and development may be the desire and skills to tinker and embody (or institutionalise) to solve problems. In development in particular, they are far more important than the ability to ideate and innovate. Persistence and not mutation is what drives development (and much else in life)… It’s therefore apposite that development embraces and elevates the attributes, skills, and values of problem-solving through the process of tinkering, embodying, iterating, and scaling, instead of the current fetish with new ideas and innovation.

A framework for public funding of innovation and startups

I blogged here exploring models of innovation funding generates the greatest bang for the buck in terms of achieving the primary objective of catalysing innovation. This post will provide some analytical frameworks to formulate a policy on innovation funding. This (on the importance of portfolio management activities), this (on an industrial policy for funding startup innovation largely through grants), and this (on the success of Maharashtra’s Defence and Aerospace Fund) are other recent blogs on the topic. This post will summarise all the takeaways and outline some guidance on startup innovation funding. 

The policy objective of startup financing is fundamentally to expand the envelope of investible startups and innovations and thereby crowd-in private risk capital. What is the best approach to achieve this objective?

Answering the question requires addressing the challenge of whether public innovation funding primarily expands the investable universe (genuine additionality) or primarily subsidises returns on investments that would have happened anyway (return-amplification / crowding-in of already-attracted capital). 

In the context of infrastructure, I have written here arguing that India’s efforts to crowd-in private capital into infrastructure sectors through the likes of IIFCL and NIIF, and IDFC earlier, have struggled to deliver the additionality (in terms of derisking sectors outside the traditional strongholds of public private partnerships) as the new institutions have ended up competing with the private sector for investments. 

What does the empirical evidence on these efforts globally report? 

A useful framework for thinking about this would be to categorise funding into three buckets: seed/angel stage (one that leads to proof of concept and lab validation, TRL 2-4), technology/product development stage (includes prototypes and pilots, TRL 4-7), and commercial scaling stage (TRL 7-9). The first category is pure incubation of ideas through grants; the second is about expanding the pool of scalable innovations; and the third is about derisking and crowding in private capital to scale innovations. 

The first stage, being the riskiest, will have the greatest additionality from public funding. It is invariably grant-funded, and gets the biggest share of public funding focus across countries, also because it is essential to create the pipeline of startups that can be feedstock for VCs and other investors. It is no good to have a VC ecosystem without a strong investible startup pipeline in the prioritised technologies. 

In the second stage, being the “innovation valley of death”, grants may be the best option. While instruments such as a Simple Agreement for Future Equity (SAFE), popularised by Y Combinator, and other forms of convertible funding are commonly discussed in the context of technology/product development, the evidence from successful global cases points to grants, with at best hybrid forms like clawbacks or profit sharing. Interestingly, apart from India (BIRAC and MEITY MSH), no major country uses SAFE in public funding.

This is because while investors obviously favour equity instruments like SAFE in pure private market contexts, they create problems with the determination of future cap tables, significantly diluting entrepreneurs and diminishing their incentives at so early a stage of the startup’s journey, and also making them significantly unattractive for commercial investors (who generally prefer startups without the encumbrances from public shareholding). 

It can be observed that those with risk capital instruments tend to kick in only at the TRL 6-7 stages. 

In this context, it is worth briefly discussing the critiques of grant funding to startups. Those who critique giving grants to startups do not realise the central role of public funding in deepening the innovation ecosystem for commercial capital to then come in. In countries like India, where early-stage risk capital is tiny, public funding is critical to create a deep and broad pipeline of innovations. The concern of possible incentive distortions from giving away free money is largely minimised through milestone-based tranches or conditional grants. Critics also tend to conflate these two categories of funding with the third stage of commercial scaling capital, which we now turn to.

The dilemma between expanding the pool of capital and returns-amplification is most relevant to this stage of commercial scaling capital. The global evidence on this is mixed. The Israeli Yozma program, which deployed funds through Fund of Funds (FoFs), is thought to have catalysed the country’s vibrant VC industry. 

However, other examples point to returns-amplification. A study of IFC’s blended finance deals in the 2000-20 period finds comparable financial returns to non-blended projects but “no statistically significant excess private mobilisation beyond what IFC’s standard lending would attract.” In other words, blending did not increase the quantum of private investment — it redistributed risk between IFC and private co-investors. 

This finding is echoed in a 2023 study of SIDBI’s Fund of Funds for Startups (FFS) 1.0 by the Impact and Policy Research Institute (IMPRI), India’s Startup Engine: A Policy Review of the Fund of Funds Initiative. It finds that most FFS 1.0 capital went to established VC funds that would have raised capital independently. It also found “crowding-in effect was primarily reputation/signal, not financial additionality,” the government’s involvement functioned more as a validation of fund managers to private investors than as a necessary injection of capital. While FFS 1.0 delivered a 2x mobilisation ratio, it was mostly in already-functioning VC markets.

This brings us to the question of the most cost-effective approach to achieve the public finance objective while supporting commercial scaling. The options span the spectrum from directly investing in the startup to indirectly investing through FoFs. 

While the former allows for targeting the riskiest innovations/startups, it creates the challenge of due diligence, which can be addressed through co-investment with professional investors that piggyback on their diligence. While the latter limits the control over who/what is funded, it allows full play for professional investment practices. 

In either case, the nature of the entity that deploys the public funds is important. A fully public or majority public shareholding corporation, whether non-profit or for-profit, will struggle to deploy risk capital and will be hobbled by the restraints of the General Financial Rules (GFR) and the vigilance from oversight agencies. It is for this reason that there is no instance from India of a government-owned entity directly making risk capital investments (apart from the Maharashtra Defence and Aerospace Fund). Its alternative, a majority privately owned entity or a Category I Alternative Investment Fund (AIF) with private Limited Partners (LPs), cannot avoid the returns-amplification problem. 

In the circumstances, the most prudent and effective strategy would be the FoFs route with some sharply defined conditionalities. The funds could be committed at concessional terms - subordinate equity, first loss buffer to a certain threshold, capped returns, warrants with lower liquidation preference, etc. It should be complemented by broad mandates on the nature of investments made, specifically on the TRL stages of the innovations, and pre-defined technology areas. 

When public capital is subordinated to private capital in the waterfall through any of the aforesaid approaches, the public subsidy is targeted precisely at the risk premium that blocks private investment. Return-amplification is minimised because private investors bear disproportionate upside — they are not getting a free subsidy on already-viable deals.

In this context, the RDIF is instructive. For a start, all its funding is debt or equity and only for TRL 4 and above stages. It has three modes of investing based on where it stands in the returns waterfall. In the first mode, RDIF effectively absorbs the first losses and receives distributions after private investors have received their hurdle rate. In the second mode, it receives distributions pari passu with other contributors at the same hurdle rate and IRR. In the third mode, it receives distributions at a higher priority or higher IRR than private contributors. 

While the first mode is a good example of concessional lending as discussed above, the second and third modes may need to be justified on other considerations. Scarce public capital should flow to those areas where it has the highest additionality. While it prescribes the broad areas of investing, it may not suffice in pre-empting returns-amplification investing. 

In the circumstances, the RDIF runs the risk of ending up with the same problems as those with the likes of IDFC and NIIF in infrastructure (whose portfolios clearly indicate that they tend to compete and crowd-out rather than crowd-in private capital). It may struggle to realise the promised additionality. For instance, it is most likely that most of the funding will flow into the TRL 8-9 innovators in the less risky among the defined areas. Finally, I’m not sure how Focused Research Organisations (FRO) can deploy returnable capital in startups, unless they merely act as pass-throughs to FoFs. 

If the second-level fund managers (SLFMs) of RDIF are required to meet additionality criteria (invest in TRL 6-9 companies they would not otherwise fund; report on portfolio-level additionality; face consequences for drift towards safe/commercial deals), the public mandate will be preserved. But without this discipline, every SLFM will cherry-pick the best deals, and the public capital risks becoming a subsidy for private returns. At best, public capital ends up competing with private capital and marginally expanding the large enough and growing pool of risk capital. 

Finally, the biggest constraint to scaling is finding the deployment platform in a country where the indigenous product ecosystem, especially domestic OEMs, is very limited. In the circumstances, public policy must play an important role in value addition by facilitating the creation of scaling pathways. This could be through direct procurements (solar cells, smart meters, street lighting LEDs, etc.) or domestic content mandates (cameras, telecom equipment, etc.). This has been a very important pathway for commercial scaling in both the advanced countries and in China, but it will be a challenge for India’s public policy. It is also for this reason that investors should pursue proactive portfolio management in terms of actively facilitating the linking of startups with the public procurement pathways. 

In conclusion, a few points to be borne in mind. One, grants at Stage 1 (TRL 1-4) are the highest-additionality instrument globally. No other instrument produces a comparable expansion of the investable universe. The evidence is unambiguous. Two, since private capital will remain scarce, public capital is critical to develop the pipeline of risky innovations and startups. Three, this nature of funding and additionality will also largely apply to the stage of technology/product development.

Four, a blended fund with a derisking public tranche and a set of sharply defined target investment-related conditions, is the highest-additionality structured instrument for commercial scaling. The public tranche absorbs the risk premium, and private capital fills behind. Five, government procurement is the highest-additionality instrument for commercial scaling for hardware companies. Procurement creates more private capital crowding-in than any equity instrument, because it proves market demand.

Finally, as public policy interventions to realise the aforesaid objectives, there are perhaps two low-hanging fruits. One, there should be a portal that consolidates all the startups financed by state and central government departments, and it should become the primary universe of the pipeline for risk capital funding. This portal should be tightly integrated with the ecosystems of VCs and other investors. Second, there should be active portfolio management at the level of all departmental funds to facilitate access to larger public risk capital funds like RDIF and SIDBI FFS 2.0. The objective should be to ensure that promising publicly financed innovations do not remain stranded.

Weekend reading links

1. Power constraints could emerge as the biggest bottleneck to America's AI growth vis-a-vis China.

Beijing has already prepared by installing an eye-popping 1,500 gigawatts of new energy capacity since 2021, taking its total to 3,891GW. However, the US has not: its installed capacity has barely risen in recent years, and now sits around 1,373GW — or less than what China added in just four years. This is shocking. Worse, China will add over 3.4 terawatts of electricity-generation capacity in the next five years, according to Bloomberg — six times as much as the US.

2. Industrial policy and infrastructure development are back again as priorities for international development actors in Africa, after couple of decades of dalliance with RCTs and small interventions. 

3. The declining labour productivity growth rates.

Between 1950 and 1973, the metric, based on output per hour worked, rose at an annual average of 4 per cent across developed economies. But the rate halved to 1.9 per cent from 1973 to 2009. And since the financial crisis, it has slowed further, averaging just 1.2 per cent between 2009 and 2025.

Pension systems are grappling with increasing dependency ratios.

When the German retirement age was set at 65 in the 1910s, life expectancy was below 50. It has now increased to over 81, while the retirement age is only set to increase to 67... In the 1960s, Japan had eight or nine people aged 20 to 64 for every person aged over 65. Now it is just over one.

4. Richard Hass makes the point that since America chose the war, it must also make the choice on ending it. This is an important point.

America did have other viable options, above all diplomacy, especially as no convincing case has been made that an imminent threat had to be dealt with militarily. The contrast between Washington’s near-unlimited willingness to compromise and demonstrate patience when it comes to persuading Russia to end its aggression against Ukraine and its unrealistic demands and lack of patience with Iran in the run-up to this war is as stark as it is telling. Ukraine’s offer to help defend against Iranian drones while Russia reportedly provides intelligence to Iran only makes the double standard worse.

5. Around 14.5 million barrels of oil transit the Strait of Hormuz daily. It has shrunk rapidly.

More on the Strait

Iran’s ace in the hole has been its de facto blockade of the strait through which one-fifth of the world’s oil and liquefied gas normally flows. At its narrowest point, the strait is less than 21 nautical miles wide, putting tankers perilously close to drones and missiles from Iran’s southern coastline. Tehran now has near-total sway over the Gulf oil market, forcing neighbours such as Iraq to almost entirely stop production and trapping roughly 300mn barrels of oil and gas in the region, a number that rises by about 20mn every day...

With its new supreme leader, Ayatollah Mojtaba Khamenei, announcing his goal to keep the strait closed indefinitely, Iran has wrongfooted oil traders who had always presumed that US military might would keep the waterway open. Iran has never blocked the strait before, despite its previous threats.

6. Some facts about Indian equity markets.

If we look at data from January 1, 2025, to the end of February 2026... emerging market (EM) equities were up 51.4 per cent, international equities were up by 47 per cent, the United States rose by 18.4 per cent, and total world equity returns were 28.3 per cent. In contrast, over this same 14 months period, India was down 0.7 per cent, the second-worst performing market in the world, with only Saudi Arabia performing worse. In fact, India and Saudi Arabia are the only two markets that are actually down (all returns in US dollar). This is when Korea has tripled, Brazil is up 80 per cent, and Taiwan has risen by more than 50 per cent. We have underperformed the EM benchmark by 5,000 basis points in just 14 months... Absolute foreign ownership of the Indian market is at a 15-year low, and we see foreign portfolio investors (FPIs) selling on a daily basis. India has received zero net foreign flows over the last five years, a very long time indeed.

It has definitively debunked the There is no alternative (TINA) hypothesis.

Our markets had done very well, and many other large EM countries were seen as uninvestable. We are the fastest-growing economy in the world — where else will FPIs go? This was the narrative. This myth has been debunked. If they wish, FPIs can totally ignore us. Five years of net zero flows. There are always choices for capital, and capital only chases potential returns. If we do not offer a good risk/return proposition, nobody will come.

7. Middle East has hundreds of desalination plants.

8. On Monday, the benchmark Brent crude price surged to $119 a barrel before diving to $84, the biggest intraday swing in dollar terms ever. 

9. The Government of India employee count (incl Railways) has remained stationary for the last decade. 

10. M Govinda Rao on the 16th Finance Commission. 

11. After starting out importing everything from China, Ukraine can now make drones with no components imported from China. 

Ukraine will not be mass-producing drones with no Chinese components anytime soon, because it’s still much cheaper to use them. Given China’s dominance of global manufacturing, it is hard to define any drone as truly “China-free.” Many components made outside China still contain Chinese parts or raw materials... Ukraine is one of many nations that have been working to reduce their reliance on Chinese supply chains... Two companies in Ukraine that have built “China-free” drones were picked to compete for contracts in a Pentagon “drone dominance program” under which the United States plans to buy thousands of low-cost attack drones. One of the companies, Ukrainian Defense Drones Tech Corporation, where the men were soldering circuit boards in the basement workshop, was among 11 in all selected last week for possible American drone orders... Ukrainian Defense Drones began making drones in 2023. Initially, all of its components were Chinese. Within a year, however, it had localized production of carbon fiber frames and antennas. By 2025, Ukrainian Defense Drones had expanded to produce flight controllers, speed regulators, radio modems and video transmission systems. Essentially, all its components were made in Ukraine except for the cameras. The company has since gained technology for cameras, too, which it hopes to produce in Europe... 

In the first year after the Russian invasion in February 2022, nearly all of Ukraine’s drones came from China. As demand surged, Beijing imposed export restrictions in 2023 and expanded them in 2024... As the rules tightened, Ukraine resorted to middlemen to buy some parts, and Ukrainian companies began to view the Chinese market as increasingly unreliable. Kyiv turned its focus to building its own drones, and eventually to doing so with fewer Chinese components. By 2024, the vast majority of drones that Ukraine sent to the front were assembled domestically — but still almost entirely with Chinese components. A year later, however, the share of parts from China in Ukraine’s drones had fallen to about 38 percent... Ukraine still buys cheaper Chinese components because the Ukrainian military needs huge numbers of drones and has a limited budget to buy them. Drone missions fail at very high rates, another reason that Ukraine tries to keep costs down.

12.  Finally, on how AI has impacted the Iran-US/Israel war. 

AI is reshaping how the US military makes decisions in war — a shift clear in Iran, where the Pentagon says it struck more than 2,000 targets in just four days... “If we look at the campaign against Isis, the coalition struck around 2,000 targets in the first six months of the campaign in Iraq and Syria,” said Jessica Dorsey, who researches the use of AI and international humanitarian law at Utrecht University... The unprecedented tempo of targeted attacks has been driven in part by AI systems that sift the torrents of intelligence data from drones, satellites and other sensors, generating strike options far faster than traditional human-led planning. The conflict also marks the first battlefield use of “frontier” generative AI models... helping commanders interpret data, plan operations and provide real-time feedback during combat. Over the past two years, the US Department of Defense has extensively integrated AI-enabled technology within its operations. The primary operating system for the Pentagon’s data is Palantir’s Maven Smart System, which alongside Anthropic’s Claude model forms a real-time data analysis dashboard for operations in Iran... 

During a live military operation such as Operation Epic Fury in Iran, Palantir’s Maven platform acts as the software “brain”. It supports the entire so-called kill chain — finding and hitting a target during active conflict. That ranges from identifying and prioritising the target to selecting the appropriate weapon and finally assessing the battle damage. Traditionally, kill chains involved printing off documents and waiting for a senior commander to study and approve it. “Those [older] kill chains are measured in hours and sometimes days,” said a defence tech expert who asked to remain anonymous. “The point of [AI] is to shrink that into seconds and minutes, almost instantaneous.”... As of May 2025, the Maven system was used by more than 20,000 users across 35 military entities in the field, according to public comments by Vice Admiral Frank Whitworth, director of the National Geospatial-Intelligence Agency. That number may be closer to 50,000 users in the US today, according to defence researchers, with Nato also signing up to use Maven in 2025...

The bombing of a girls’ primary school in Minab, in southern Iran, further illustrates the lethal risks of quickly generated or improperly vetted targets... In Iran, AI has potentially already been involved in identifying exponentially more targets than in previous wars, said Utrecht University’s Dorsey. Those targets could have existed beforehand — or they could have been generated quickly by AI systems, creating a serious concern about how carefully these have been vetted as required by law, she said.

This about the Minab school bombing.