Constraints to large manufacturing facilities in India

The conventional wisdom on firm size in India is that labour regulations on hiring and firing (specifically the Industrial Disputes Act, IDA) hinder its growth. I have blogged here and here pointing to research that raises doubts about this argument and here pointing to the importance of the lack of entrepreneurial appetite and dynamism as a contributor. My co-authored paper here dwelt extensively on the problem of firm size and productivity. This post will discuss the issue. 

Abhishek Anand, Arvind Subramanian, and Naveen Thomas have a paper with some useful quantitative insights on firm-size. They show that the real problem is not the IDA, large plants are more productive, and that there are very few large globally competitive plants. These are already well-known facts, though not the reasons and the paper does little to explain them. But the contribution of the paper is to draw attention to a flaw on how the ASI data on firm size is captured which might be overestimating the firm size distribution in India. 

For documentation of the numbers, I’ll extract some of their findings. These are the three headline findings.

First, that this phenomenon of multi-plants has been growing over time and is quantitatively significant, accounting today for over 25.16 percent of total employment in all plants and 35.48 percent of employment in large plants. Second, ignoring it leads to over-stating the change in the size of plants since the early 2000s. It is popularly believed that Indian plants have become larger but we show that that is not the case and on some metrics large plants may have even become smaller… Third, that the multi-plant phenomenon seems to be a response to some underlying friction. It seems to be an endogenous device for Indian capital to keep their operations small, presumably as a way of coping with the regulatory burdens and risks imposed not, or not just, by labor laws but the broader political environment, shaping capital-labor relations.

Some other findings

The contract labour phenomenon is much greater—almost twice as large—in single plants compared to multi-plants… Single plants are both more “productive” than multi plants for any given level of employment and this differential increases as employment size increases. For example, at employment of 200 workers, single plants are about 9 percent more productive and at 1000 worker plant size, that wedge increases to over 21 percent… even in 2022, the 75th percentile plant in India employed about 53 workers, and the 90th percentile firm employed 128 workers. These cannot be considered large by any means…

Bangladeshi plants export on average 95 percent of their output compared to 37 percent for India… if we compare plants size in India and Bangladesh, we find that Bangladeshi plants are consistently bigger at every threshold with the size differential rising as we go to higher thresholds for size: for example, the 95th percentile firm in Bangladesh is about 40 percent larger than its Indian counterpart… Plants employing more than 200 workers account for roughly 85 percent of all employment in Bangladesh. In India, the comparable correct number is 50 percent. And plants employing more than 1000 workers account for roughly 41 percent of all employment of Bangladeshi plants. In India, the comparable correct number is 15 percent.

They also write on labour regulations

Much of the literature on employment size has focused on the Industrial Disputes Act (IDA) and on examining whether the thresholds in them have shaped plant size. This has been a distraction, impeding our understanding of plants size and the labor market… We find that there is a clear tendency for plants to remain small even beyond the 100 worker threshold set in the IDA so that it is not the law per se (which is after all legally irrelevant beyond the threshold) but other factors that shape firm behaviour.

The authors also discusses how labour hiring and management decisions may interact with the firm size, but different from the conventional wisdom on the restrictions of the IDA, in a two-part oped here and here.

Rising contractualisation of labour — from about 22 to 41 per cent over the first two decades of this century — has been an important response of Indian firms and their management to the regulatory environment. Firms such as TeamLease act as brokers, taking upon themselves the burden of complying with labour laws so that manufacturing firms themselves do not have to. But somewhat puzzlingly, we find that contractualisation is lower in labour-intensive industries than in non-labour intensive ones. This is puzzling because if contractualisation is a response to labour laws and their burdens, the incentives to do so should be greater in labour-intensive industries. For example, in labour-intensive industries, the share has risen from about 23 per cent to 31 per cent over two decades, but in other industries from 19 per cent to 47 per cent. Unpacking this further, we find that recourse to contractual labour is greater in single plants than multi-plants. We also find that at the margin, the incentive to substitute contract labour for full-time employees rises with employment size in single plants but does not do so uniformly for multi-plant units.

One explanation is simply that in multi-plant units, flexibility in hiring and firing labour comes from the fact of having many plants. In single plants, there is no such flexibility, which renders the use of contract labour more important. We were told by the CEO of a large exporting firm that in the event of, say, a drop in orders from one client that affects one plant, the firm can redeploy labour in another plant without having to terminate their employment, which would be the only option in a single plant establishment. In other words, multi-plants and contract labour are both devices that increase flexibility but in different ways and for different situations and work as substitutes. According to the CEO, it would be more competitive internationally if its plant sizes could be greater. But it chooses not to grow as a matter of diversifying policy and legal risks and because of onerous regulations. 

The risks may not be the law per se but stem from the broader political environment in which the firm feels it would be vulnerable to the whims of the Centre and state governments, and also to labour in the event of frictions or disputes. A dispute in a big plant would entail greater risks relative to that in a smaller plant: In an extreme situation shutting down a plant with 500 employees is less costly than one with 5,000 employees… The constraints imposed by the thresholds in the Industrial Disputes Act are not the only deterrents to scaling up; it is the pervasive uncertainty in the business and regulatory environment that seems to compel firms to fragment their operations regardless of scale.

The conventional wisdom on firm growth in India is that they are constrained by labour laws that discourage expansion beyond certain employment levels. While there’s some truth in this, there are other (perhaps more difficult to overcome) glass ceilings that Indian firms encounter in their growth trajectories, that go beyond even the political and regulatory risks imposed by labour size alone in a plant.

While the paper discussed above goes beyond the narrow instrumentality of the labour hiring and firing constraints posed by the IDA, and points to the uncertainties in the country’s political and regulatory environments that deter labour expansion, the problems may go much deeper. 

Large size is not just about labour. Let’s explore this a bit more. A few questions are in order.

Is it the case that instead of some unique Indian contextual reasons, the very nature of the political and regulatory risks of large plant size by itself, irrespective of the country but especially so in developing countries, are too onerous as to discourage firms in general from establishing large plants? Is it possible that the large manufacturing plants, of the kind we are interested in, emerge only through special circumstances (and not through the general market dynamics of demand, competition and firm growth)? Is it possible that irrespective of the general political and regulatory environment, such large-sized factories can emerge only through proactive support from the government? Is it the case that governments in India while not explicitly against large size have not been active promoters of large-sized plants in particular and large firms in general, preferring instead to support small and medium-sized firms?

There’s nothing automatic or market-based about the growth of successful firms from small to medium, and from medium to large, especially in the manufacturing sector. Firms hit binding statutory, economic, and entrepreneurial constraints when they reach a certain size. For example, as I blogged here and here, there might be significant risk-appetite bounds to be crossed in such transitions, which many, if not most, successful entrepreneurs struggle with.

Apart from the political and regulatory uncertainties relating to labour discussed above, large size is associated with generally more onerous compliances, firm growth requires adding new factories, hiring good professional executives besides expanding the labour force, revisiting ownership holding and corporate form, changing management structure and practices, tweaking business models, acquiring new customers, assuming more debt, etc. Firm size also invites greater external scrutiny of all kinds in general. Each of these imposes significant requirements on the owners and the management, including decisions that require overcoming entrenched norms. 

For example, many successful family-owned firms struggle to trust professional executives and prefer to avoid assuming more debt. The smaller size of the enterprise helps them control their business without relying on outsiders and professional managers. It also limits the commercial risk exposure from a business downturn. They can rely on some local and loyal staff to manage the business. Scaling also requires investments in greater automation and adherence to greater standards, which demands capital and debt. All this creates the danger of losing control and assumption of greater business risks.

Further, as firms become large, they must compete with their more aggressive national and foreign competitors to acquire and retain customers and iterate continuously to improve their products, business lines, and delivery models. Any growth that involves expansion into foreign markets demands continuous productivity improvements, agility, and a high risk appetite. The firm cannot stay static and must be dynamic in all aspects. All this requires a growth mindset - an appetite to assume risk and an ambition to expand and become an industry leader. However, most Indian firms appear happy and content with their current market share, at best growing marginally, and strive only to retain it. The low R&D investments of even the leading corporates in the country are a reflection of this mindset. 

The growth bounds are perhaps amplified for Indian entrepreneurs with their struggles of doing business in environments that are sometimes downright hostile and mostly not-so-easy. These constraints manifest in the general inability of large numbers of very good small and medium enterprises to break out and emerge as large-scale firms.

So what are the policy takeaways to address this complex and binding constraint?

A headline takeaway that should be strongly internalised among policymakers in India is that scale manufacturing does not emerge on its own and requires government support. This is important since the current guidances and norms are not only to support SMEs but also to avoid supporting larger firms. There are some other misplaced but widely held beliefs, amplified also by the fear of vigilance agencies - once the SME grows the industrial policy support should cease, and firms once supported must not benefit again etc. 

It must be noted that government support has been central to the emergence of scale manufacturing in China and all the North East Asian economies. The zaibatsus and chaebols of Japan and South Korea respectively created the culture of world-class scale manufacturing in those countries, with considerable support from their governments. 

The likes of Morris Cheung of TSMC and Grace Wong of Luxshare Precision Industry got extraordinary levels of support from the governments of Taiwan and China. Thanks to the support the Chinese government provided to help her company emerge as a leader, the latter has in a little over a decade emerged to become Apple’s second biggest supplier, coming only behind her previous employer Foxconn. The government virtually forced Apple to enlist and grow Luxshare as a contract manufacturer for Apple products. The Chinese government has focused on the creation of large-scale manufacturing giants, especially but not only among state-owned enterprises. 

A general feature of the successful North East Asian industrial policy has been to weed out the weak and double down with support for the strong firms and let them grow in size.

A big positive about India’s scale manufacturing strategy is the arrival of large foreign contract manufacturers and domestic corporate groups like Tata. These were perhaps the only ways scale manufacturing could have emerged in India. Tata has taken the lead on electronics manufacturing in India, with active government encouragement and PLI and other facilitation support by central and state governments. It’s incumbent on the other large conglomerates with manufacturing legacy to step up and emulate their East Asian counterparts and establish scale manufacturing facilities.

It would now be interesting to closely study at least three possible future trends over the coming 5-10 years. One, are these large contract manufacturers spawning a large and growing ecosystem of component manufacturers and sub-assemblers in India? Two, is the initial cohort of contract manufacturers engendering emulation by domestic corporates and entrepreneurs, especially those with existing medium-sized firms in the industry? Three, is the presence of large contract manufacturers resulting in knowledge and technology spillovers across the regional economies and contributing to productivity improvements? 

I can foresee one important problem with the realisation of these objectives. The aforesaid trends require significant AND increasing value addition by the contract manufacturers. It provides the market signals of market and value-addition growth to both foreign component manufacturers and those dynamic and ambitious domestic entrepreneurs to set up shop. Firms look for growth both at the extensive (volume of the same business) and intensive (increasing value addition and shift to adjacent market segments) while making long-term investment decisions. It’s possible that the contract manufacturers, especially but not only the foreign ones, could largely remain stuck in the lower-value assembly stage and not have the incentives to continually move up the value chain. 

It’s therefore important that public policy closely watch the emerging trends and engage to achieve the objectives above. The current success in mobile phone assembly should not blind us to the need to quickly move up the value chain by doing more component manufacturing domestically. 

The PLI scheme is a good place to start. The next round of PLI-scheme incentives could be linked to domestic value addition as against sales (even with all the challenges of its accurate measurement). It may even be useful to consider extending the incentives on the existing PLI scheme beneficiary firms for another five years, but this time on the incremental value addition over a baseline. There should not be any limits on the number of times a firm can access PLI scheme benefits (in the different scheme rounds). 

These and other policy interventions might be necessary to achieve the objectives underlying the three aforesaid desirable trends.

Thoughts on affordable housing - VIII

The provision of affordable housing requires policy action at multiple levels to reduce land and construction costs and increase supply. This post proposes a policy mix for state governments in India to lower housing costs in their urban centres. 

Historically central and state governments have pursued a siloed approach to housing. The government's focus has been to increase supply by providing free or concessional land, subsidies, and/or interest subsidies to individual households. There are hard fiscal limits to such efforts. A few state and urban local governments have occasionally tried some very limited schemes involving urban planning relaxations. 

However, there has been no comprehensive scheme that brings together all possible policy levers to reduce costs and expand supply in a manner that genuinely promotes affordable housing sustainably. 

Since the land extents are fixed and limited, meaningful expansion of housing supply can happen only by relaxing floor area ratios (FARs) and going up vertically. Since statutory charges and taxes on land and construction form about 40-50% of the total construction cost, any effort at expanding the supply of affordable housing requires reducing them significantly. Even with all these enablers, there will be market and co-ordination failures for several reasons. Market failures will require some light touch and simple regulation. Coordination failure requires proactive engagement with builders and ease of doing business in the industry. Finally, all these must be complemented by facilitating households’ access to bank loans, thereby creating a deep mortgage finance market. 

Fortunately, state and city governments in India have the policy levers and institutional mechanisms to engage on the problem holistically. 

The instruments available in theory to the government to encourage affordable housing are the FAR, land registration fees, statutory charges (layout development fees, betterment fees, building permission fees), property tax, linkage with bank loans and the interest subvention subsidy under the Credit Linked Subsidy Scheme (CLSS) (3%-6.5% for MIG/LIG/EWS), and land. 

Governments need to provide significant enough relaxations on the building regulations to at least partially offset the high land cost and lower the total cost of affordable housing construction. This can come in the form of higher FAR and height limits, lower setbacks, and relaxations on minimum plot and road-width sizes (the current sizes are very large in many states). These relaxations allow developers to provide more units on the same land extent. 

In this context, a landmark reform would be to adopt a policy that associates a building right (in terms of FAR) with a property ownership right. The landowner would have to buy the additional FAR beyond the building right, to the total FAR extent permitted by the Master Plan in that area. I have co-written here on this. This policy allows governments to use FAR as an instrument of revenue mobilisation and for formulating policies like those on transportation and affordable housing. 

In this context, it’s to be borne in mind that Mumbai with a base FAR of 1-1.33 and purchasable FAR of 0.67-1, raises nearly Rs 6000 Cr each year from outright sales of FAR. In the new Andhra Pradesh capital city region, Amaravati, section 210.5 of AMRDA Zoning Regulations says that “wherever FSI exceeds 1.75, the applicant has to pay impact fee as levied by APCRDA from time to time”. 

The state government could come up with an affordable housing scheme that would either directly offer or enable access to the following

1. Higher FAR, either free or at low purchasable rates

2. Reduction in land registration fees

3. Reduction in statutory charges

4. Lower property taxes for the first 5-10 years.

5. Linkage with banks to avail of housing loans and the interest subvention subsidy under the CLSS. 

The exact details or extent of the benefits given under each instrument should be worked out. It should be left to the city governments to tweak the extent of benefits but with an upper limit on how much benefit can be given (if only to provide guidance and ensure some discipline among them).

Public land at concessional rates should be used only as a last resort. Given the limited extent of such lands, it’s not a sustainable policy to generate an affordable housing supply.

The Real Estate Regulation Authority (RERA) provides the ideal institutional platform for the implementation of such projects. The state government could establish its own applications portal that links the RERA portal, and the websites of the municipal governments and banks by using APIs to facilitate seamless processing of applications to avail benefits under the scheme.

The scheme could be two-tiered – one with more generous benefits for housing units up to 600 sq ft and slightly less so for those up to 900 sq ft. Some light-touch income qualifications could be kept to eliminate egregious abuse. 

The RERA could fix a reasonable city-wide upper limit for each category. This rate could be the same for all cities, or cities could be grouped into 2-3 categories and a rate fixed for each or determined separately for each city (for the larger ones). While some might be considered too invasive, it’s common across the utilities sector. In housing too, Singapore’s lessee-occupied public housing is a good example of a cost-plus-regulated housing market. And the very nature of the housing market dictates that it cannot avoid some form of price regulation. 

The state government could announce an affordable housing scheme that offers all these benefits for the buyers of RERA-registered affordable housing apartments. The housing departments/units of the state and municipal government could work together to facilitate apartment buyers’ access to bank loans and the CLSS subsidy. The state and local governments could work with banks to get them on board the scheme and mitigate any concerns arising from previous bad experiences with such governments. The Government of India should consider expanding and extending the duration of the credit-linked subsidy scheme. 

This would be one of the less fiscally burdensome (at least in terms of current fiscal expenditure, though not in terms of revenues foregone) policies possible to meaningfully address a problem like the affordable housing shortage. 

To make it politically attractive, all the benefits would be wrapped together in one bundle, their monetary value captured, and delivered as a package to all affordable housing beneficiaries. It’s also essential to ensure that there’s no political economy-related moral hazard among mortgage holders from the government’s involvement in the scheme. One way to mitigate it is to avoid any direct involvement between the government and the apartment buyers, and confine government facilitation to developers and banks. 

The challenge will be to get builders and developers to pass through the benefits of this package to buyers. In markets where the demand for such housing far outstrips supply by orders of magnitude and the market itself is prone to and rife with abusive practices, some amount of regulation in this regard is essential. Over time, if the scheme creates its supply, there will be pass-through at a steady state. 

The administration of a scheme like this requires good-quality data. There should be baseline data on the construction costs and market prices of both affordable and regular housing. The value of the benefits provided should be monetised as reference data (the applications portal could show this value for each applicant). This data should be collected periodically, analysed, and used appropriately for decision-support on the scheme. This analysis will help determine the extent of pass-through in affordable housing.

Weekend reading links

1. NYT has a story on how Heidelberg Materials, a Germany company, is trying to produce cement in Brevik in Norway, without carbon emissions by using carbon capture technology.

More than half a ton of the gas arises from every ton of cement that a plant like this turns out. Early next year, carbon dioxide from the facility will be chilled to a liquid, loaded onto ships and carried to a terminal near the city of Bergen, farther up the Norwegian coast. From there, it will be pumped about 70 miles offshore into rocks a mile and a half below the bottom of the North Sea... It helps that the Norwegian government is underwriting 85 percent of the up to 400-million-euro cost of what will be the first large, commercial-scale effort to strip carbon dioxide from cement and bury it... Hasan Muslemani, head of carbon management research at the Oxford Institute for Energy Studies, estimated that the cost of capturing the carbon dioxide would be up to double current cement prices. Heidelberg is counting on customers to be willing to pay more for a product that comes with green credentials...

Mr. Houg suggested during a presentation to journalists that the carbon-free cement might cost three times the market price for ordinary cement, around €130 a metric ton... Cement production, which accounts for nearly 7 percent of energy-related emissions, presents one of the knottiest problems for emissions reduction... Heidelberg estimates that cement accounts for only up to 2 percent of the expense of a large building project but for perhaps 50 percent of the emissions, so using carbon-free cement is a relatively inexpensive means of cutting emissions... Heidelberg says that when Brevik is up and running, it will reduce emissions by around half, or about 400,000 metric tons a year. That amount may seem like a big number, but it is a small portion of the 61 million tons that the company’s cement business produces.

2. Oren Cass has a provocative response to the critique of Trump's proposal to raise tariffs to level the playing field. 

In Economics, the industry-defining textbook first published in 1948, the Nobel laureate Paul Samuelson argued aggressively for free trade. He did not, however, deny that tariffs work; under the heading “Beggar-Thy-Neighbor Policies,” he listed the many ways that policies like “protective tariffs” could help “create a favorable balance of trade.” Rather, Samuelson urged that “any intelligent person who agrees that the United States must play an important role in the postwar international world will strongly oppose the above policies,” because to do otherwise would be to “attempt to snatch prosperity for ourselves at the expense of the rest of the world.” As C. Fred Bergsten, the founding director of the Peterson Institute, acknowledged in 1971, “The economic argument was always marginal” for free trade. “It was the foreign policy case which provided the real impetus for liberal trade policies in the United States in the postwar period.”

3. Ajay Srivastava points to an important trend with India's manufacturing and exports.

In FY14, India’s GDP was $2,010 billion, with merchandise exports at $314 billion and manufacturing accounting for 15 per cent of GDP. By FY24, GDP had grown to $3,900 billion, and exports reached $437 billion, but manufacturing’s share in GDP dropped to 13 per cent. With manufactured products making up 75 per cent of India’s trade in both periods, the share of manufacturing exports relative to manufacturing GDP fell from 78.1 per cent in FY14 to 64.6 per cent in FY24. This reflects a dual challenge: The shrinking role of manufacturing and its decreasing export contribution.

This about competitiveness compared to China

The cost of industrial electricity in India varies between $0.08 and $0.10 per Kwh, whereas in China, it’s $0.06 to $0.08. Additionally, lending rates in India average 9-10 per cent compared to 4-5 per cent in China. These cost differences make a product expensive. For instance, producing solar cells in India is 40 per cent more expensive than sourcing from China... India relies on foreign shipping companies for 90-95 per cent of its trade cargo, leading to higher freight costs and less control over shipping schedules. About 25 per cent of Indian cargo passes through hubs like Colombo and Singapore because many Indian ports lack deep waters for large ships, adding time and costs.

4. Japan softpower exports are competing with hard power ones.

In a report published in June the government gave figures on the value of exports by its content industry — a grouping that included anime, games, films and comics. In 2022, those exports were about $30bn. Pointedly, the report juxtaposed this with the $34bn exports of the steel industry and $38bn of its semiconductor sector.

This about the spectcular success of Japanese animation, or anime

There are currently an estimated 800mn fans of Japanese animation — known as anime — around the world, and key figures in the industry see those numbers rising towards a billion fairly soon. If achieved, that would be roughly equivalent to the global fan base of tennis... After years of growing ever more hungry for content, the overseas market for Japanese anime is now almost the same size as the domestic one and growing more rapidly... Unlike some other forms of entertainment, whose popularity expanded strongly during the pandemic but waned afterwards, anime consumption has continued growing past 2022. In 2023, the combined sales of Japan’s anime production companies surged almost 23 per cent from a year earlier to hit an all time record and, according to research group Teikoku Databank, are on course to break that again in 2024. Analysts at Jefferies cite industry projections that the global anime market will almost double from $31.2bn in 2023 to $60bn by 2030 because what was once a largely Japanese genre has comprehensively shifted into the mainstream culture of the US and Europe.

The Economist has a good description of anime

Anime is hand-drawn and usually two-dimensional, unlike the photorealistic 3danimation that has grown more common outside Japan. The characters often resemble hyper-caffeinated Tintins, with unusually large, expressive eyes, small noses, strange hair and easily decipherable emotions... Anime’s global growth has been made possible by its own evolution. Japanese comic books and graphic novels, collectively known as “manga”, have long provided the source material for anime. Unlike comic books in Britain and America, manga are not primarily or even mostly for children. Thousands of new manga are published every year, on virtually every subject imaginable, from pornography to reflections on war, which gives anime an inexhaustible range of sources.

But around 30 years ago, studios started producing more anime aimed at girls, such as “Sailor Moon”, that was less focused on fighting and robots and more on storylines and magic. And as the fan base got older, creators began making more sophisticated works, often with more adult themes... Around that time “Dragon Ball” and “Pokemon” were engaging a new generation of fans. Japan’s government noticed and eventually took action: in 2013 it launched an initiative called “Cool Japan”, in which it invested some ¥90bn to propel Japan’s creative industries abroad. It was a flop, because of poorly chosen investments, but that has not stopped the government from trying again: it wants to quadruple the value of Japan’s content industry by 2033. 

5. How the US is using the CHIPS Act money?

A good chunk of the chips money has gone into bolstering schools and vocational programmes in areas like upstate New York, where the US science department signed a memorandum of terms with Micron Technology, which plans to invest around $100bn in chip production over 20 years.

6. Ajay Kumar writes about the latest developments in quantum communications.

In August 13 this year, a significant milestone in quantum technology was achieved as the National Institute of Standards and Technology (NIST) announced the first three Post-Quantum Cryptography (PQC) standards. These standards open the door for the rollout of this technology in establishments worldwide, addressing concerns about quantum computers breaking current encryption protocols, and endangering the entire cyber landscape...

Two key technology trends have emerged. PQC and Quantum Key Distribution (QKD). PQC involves developing cryptographic algorithms resistant to quantum attacks, while QKD uses quantum mechanics to securely distribute encryption keys. Curiously, there is no consensus yet on which of these, or whether both, would become the norm. The world is split between the two approaches: The United States favours PQC, while China and Russia are investing heavily in QKD.

7. Is e-commerce worsening the Chinese deflation?

Because a growing portion of spending in China is happening online, the price reductions by the app and other e-commerce platforms copying its success have contributed to a deflationary downturn. About 60 percent of the country’s consumers buy through e-commerce, accounting for more than one-third of all retail spending, according to HSBC. “Pinduoduo is both the consequence and cause of deflation,” said Donald Low, a professor of practice in public policy at the Hong Kong University of Science and Technology. Founded in 2015, Pinduoduo has grown more quickly than its more established rivals, recently expanding abroad with its Temu brand. In its most recent quarter, Pinduoduo said revenue had risen 86 percent... 

Colin Huang, Pinduoduo’s founder and one of China’s richest men, has said that one of the company’s core values is not to sell cheap products, but to offer goods that customers will feel are less expensive than they should be. Earlier this year, Ms. Lin, the seller in Zhangzhou, said Pinduoduo had enrolled her in an “automated price tracking system” to allow the company to lower the price of her diapers whenever it detected similar products available for less. A few months after she opted out of the program, she discovered that the setting had been turned on again... Economists have studied the consequences of e-commerce on pricing for years. In the mid-2010s, economists started citing something called the Amazon Effect, for the influence wielded by the dominant online retailer Amazon.com to drive down prices across the web and at brick-and-mortar stores. Almost all retailers, including Amazon, track each other’s prices and then adjust their own using so-called dynamic pricing, when prices move according to market conditions...

Pinduoduo’s success has prompted its two largest rivals, Alibaba and JD.com, to join the low-price competition. Last year, Alibaba’s shopping site Taobao started a campaign to rate sellers based on how their prices compared to other e-commerce platforms, according to Chinese media. The sellers with better prices would receive more traffic and exposure for their products. JD.com, once known for selling high-end electronics, has also created a series of low-price campaigns.

8. The remarkable gold rally this year, from its sub-2000 low in February to the 35% rise till date.

10. Important facts about the global oil market

Nearly 60% of the world’s oil now comes from countries other than the cartel and its allies, up from 44% in 2019. America’s shalemen have become the biggest producers in the world by far. Brazil, Canada and Guyana have all increased their output in recent years. According to the International Energy Agency, production by non-opec countries will grow by 1.5m bpd next year.

11. The Economist points to the Trumpification of American policies, where the Republican candidate has set the agenda only.

Ms Harris’s immigration policy is to endorse the most conservative bipartisan reform proposal this century. Its provisions include shutting down asylum applications when the flow of irregular migrants is high. Her trade policy involves keeping, in modified form, most of the tariffs Mr Trump imposed in his first term. On tax, Ms Harris would keep most of the cuts Mr Trump signed in 2017 (raising rates only for those who earn over $400,000). On energy, she has become a convert to fracking and has been part of an administration that has seen America pump more oil and gas than ever before...

Mr Trump adopted a more confrontational approach to China than any recent president, even if his policies were in practice less scary than they sounded. The administration Ms Harris has been part of has been less verbally antagonistic but tougher in practice, banning technology exports to China and placing huge tariffs on imports of Chinese electric vehicles. On the Middle East, Ms Harris has not let Mr Trump outflank her on the right, despite pressure from within her own party to cut arms supplies to Israel. Nor does she seem in a hurry to revive the deal with Iran that Mr Trump tore up; this week she called the Islamist regime America’s greatest adversary.

12. Finally, the data centre rivalry between US and China in Asia.

There are four choke points: “internet exchange points” (IXPs), which let internet firms cut costs by routing traffic down the hallway rather than across the world; data centres; undersea fibre-optic cables; and telecoms firms. All are vulnerable to espionage. Tapping undersea cables has been a trick of spooks since the cold war. Landing stations are hubs for data interception. Back doors can be installed in infrastructure. America’s government has warned that Chinese presence at ixps creates “the capability to misroute traffic and, in so doing, access and/or manipulate that traffic”. Even if the data flowing through IXPs are encrypted, some experts think metadata could be exposed.

Infrastructure DFI must be a fully public entity

This post will try to explain why DFIs that have private investors will not be able to achieve national objectives on infrastructure financing. 

I have written here explaining the problems with infrastructure finance, categorised the infrastructure sector into those where projects need to be de-risked to attract commercial investors, and how DFIs can perform the de-risking role. 

Governments across the world have experimented with various DFI structures to de-risk and crowd in private capital to finance infrastructure projects. All the Western DFIs are fully government-owned entities which use different concessional financing instruments (concessional or subordinate debt, guarantees and credit enhancements, and equity) to crowd-in private capital into individual projects. In simple terms, while the project entities that the DFI invests or provides finance are privately owned, the DFI itself is fully government-owned. The DFI seeks to crowd-in private capital at the project level. All of them address the risks of capture and abuse arising from government ownership by having good governance systems. 

India has taken a different path, seeking to crowd in private capital at the DFI level itself. This has been a consistent endeavour from the IDFC in the nineties to the NIIF. The theory is that the government will supplement every rupee that the private sector brings in, thereby making both go longer, besides expanding the pool of private capital available for infrastructure finance. 

Let’s unpack this model. The private (or non-government) institutional investors in the DFI’s Asset Management Company (AMC) want to maximise returns from the investments made in the portfolio projects. This introduces a natural selection bias towards the nature of projects that the DFI will want to fund. 

The government instead seeks to maximise its additionality - de-risk and expand the envelope of projects that are amenable for commercial funding. It wants the DFI to invest in sectors like urban water supply and sewerage, solid waste management, electricity distribution, urban mass transit, energy saving projects (ESCO), health and education PPPs, etc., where private investors are deterred by the commercial unviability of the projects. 

There’s an inherent and irreconcilable conflict between these two objectives. Since money is fungible and corporate shareholders cannot have such conflicting goals, this model will struggle to satisfy either side. I cannot see how an AMC that brings together both private and public shareholders can meet the government’s objectives.

In the battle of the conflicting objectives, the private capital’s interests will invariably prevail. Or else, they will exit the entity. There’s no scenario under which the government’s additionality objective can be realised.

It can be argued that even without providing concessional capital, the DFI is de-risking projects by having the government as a co-investor. This would have been the case with infrastructure investments in India a decade and a half back and would be so even today in several countries. But in the current Indian context, such de-risking is minimal. This can be easily verified by examining the portfolio (and pipeline) of NIIF and scrutinising them for the additionality of scarce public finance in its portfolio of investments. It’ll reveal that the DFI has competed with other private investors and crowded out private capital. 

It’s possible to pretend otherwise and kick the can down the road by rationalising the DFI’s current investments as arising from a need to build a portfolio of commercially viable investments to establish its track record. This rationalisation glosses over the fact that no track record can incentivise the entity to invest in risky projects of the kind that the government has in mind. 

The only justification for government funds flowing into an institution with such a corporate structure is that of a Sovereign Wealth Fund, where the objective is to deploy the government’s reserves and surpluses and earn returns higher than from traditional risk-free Treasury Bond investments. 

This raises the question of why India has persisted with structuring DFI’s in this corporate form that’s different from the practice elsewhere. 

The primary reasons for structuring DFIs without majority government ownership and in PPP mode are to have an arms-length relationship with the government, to hire professional investment managers from the market, and to avoid the scrutiny of public oversight agencies like the Comptroller and Accountant General (CAG) and the Central Vigilance Commission (CVC). The latter two in particular have been critical determinants. Unfortunately, this reasoning introduces a critical design flaw in the corporate entity's commercial incentives, making it impossible to meet the government’s fundamental infrastructure finance objective. 

It’s, therefore, hard not to argue that such entities are a very inefficient use of scarce government financing. It provides free public capital to private investors with little in return for the government (apart from returns if the objective is to be a SWF). In the nineties and early noughties, the IDFC was found free-riding on public funds and lost the trust of the government. It’s difficult to foresee anything different with NIIF. 

The only alternative would be to embrace the tried and tested models of fully government-owned DFIs that have the flexibility to deploy a basket of differentiated instruments to de-risk and crowd in private capital. As I have blogged earlier, the government should channel all its different kinds of infrastructure finance initiatives, like viability gap funding, interest subvention schemes, capex support schemes etc., through this DFI. It should become the single point for all government initiatives that involve attracting private capital into infrastructure projects and should work with local, state and central government entities to achieve financial closure on their projects of such kind. 

Yes, this entity will be constrained by public oversight and will not be able to pay premium salaries to its investment managers. But it’s possible to meet the objectives even with these constraints if there are good due-diligence protocols and governance systems, and they are complied with seriously. That has been the practice globally. 

Some thoughts on industrial policy for startup innovation

The rise of industrial policy globally naturally raises the question of which instruments generate the greatest value for money. This post will examine the case of public funding of startups engaged in technology innovation. 

Thomas Hochman has a long read in American Affairs which compares the instruments used by the CHIPS and Science Act and the IRA Act. 

The… CHIPS and Science Act, works mostly through grants: funding awards are given directly to specific manufacturers by the Department of Commerce, with explicit terms around domestic production, intellectual property sharing, and reinvestment of profits. This has its benefits: the spending is easy to track, is highly targeted, and carries little risk of filtering out to foreign adversaries… But there are downsides to this approach. Direct government spend­ing requires administrative capacity: the federal, state, or local agency responsible for carrying out the program must solicit applications, review them, and make decisions about whom to award money. It also requires organization on the part of private sector applicants, who often lack the resources and know-how to put together grant applications in the first place, or even to gain awareness of relevant programs’ existence. 

The result is that grants tend to roll out slowly. At the time of this writing, almost two years since the passage of chips, less than $1 billion of the $52 billion appropriated for semiconductor funding has been formally awarded… Simi­larly, the $5 billion National Electric Vehicle Infrastructure pro­gram, passed through the Bipartisan Infrastructure Law in 2021, has led to exactly eight charging stations being built as of February 2024, while hundreds of other years-old federal spending programs have yet to make a single award.

The Biden administration’s second philosophy of industrial policy is exemplified by the Inflation Reduction Act (IRA). The IRA takes a climate-first approach, prioritizing emissions reductions over China competition, with the goal of deploying as much clean energy infrastructure as quickly as possible. Accordingly, the law works through “as-of-right” tax credits: companies that meet certain eligibility requirements can claim the credits, incentivizing clean energy investment while creating relatively little bureaucratic friction. The upshot is that IRA funding has reached the clean energy industry much more quickly than CHIPS funding has reached semiconductor firms. But this approach has its drawbacks as well. Whereas grant programs lack speed, tax credits lack transparency. Because as-of-right tax credits are facilitated through the tax code, they are protected by Internal Reve­nue Service confidentiality. As a result, no one—not the public, nor watchdog groups, nor even most members of Congress—knows exactly how much we’re spending, when we’re spending it, and on whom we’re spending it. 

The essay points to the economic inefficiency associated with tax credits, especially with tax credits that are allowed under the IRA Act to be transferred to unrelated third parties, tax-free. It highlights the example of Ashtrom Renewable Energy, which expects to generate around $300 million in production tax credits over ten years from its 400 MW solar project in Texas. It immediately sold its entire $300 million credits to a financial institution with sufficient tax liability to offset its taxes over ten years. 

Tax equity investors appear to be syndicating portions of their credits, and dedicated brokers and insurers are emerging to facili­tate trades. These middlemen are seeking to maximize their own returns, of course, and their cut ultimately reduces the amount of public money reaching actual projects. There’s also the simple fact that if tax credit recipients pursue transferability, then the credit is not worth its full dollar value to the seller. The upshot is that the IRA credits are being transferred at an average of about 93 cents per dollar of credit value—in other words, at a 7 percent loss in credit efficiency… Indeed, the cost of monetizing credits through the tax equity market can be as high as 15 percent of the credits’ value… Reports from intermediaries suggest that the institutions that have historically been involved in tax credit deals are the same ones buying up IRA credits today: banks, insurance companies, and other major corpo­rate taxpayers. This is not to mention the benefits to the many syndicators, tax advisers, and insurers that have emerged in the wake of the IRA’s passage, all of whom are skimming cents off the top of every dollar of credit value.

The essay concludes by pointing to the advantages of grants

Grants have been a staple of U.S. industrial policy for decades, from the Advanced Research Project Agency’s funding for energy R&D to the Department of Defense’s defense industrial base grant program. And while grants often involve more bureaucratic processes compared to tax credits, they offer several advantages. First and foremost, grants provide greater up-front control and ability to target funds strategically… grants allow for more due diligence and selection of recipients aligned with those strategic goals. Policymakers can target specific industries, technologies, or geographic areas, and put in place guardrails to prevent funds from benefiting foreign competitors.

Grants also offer enhanced transparency and accountability compared to tax credits. While tax credit data is shrouded in IRS confidentiality, grant recipient information is typically public… Third, grant agreements provide an opportunity to set specific conditions and requirements for recipients. Agencies can specify the scope of work, set milestones and reporting requirements, and include clauses to suspend or reclaim funds if conditions aren’t met… For all of these reasons, grants are particularly well-suited for advancing national security priorities and shoring up strategic industries. 

This debate has relevance for India. Apart from the production-linked incentive (PLI) scheme aimed at encouraging domestic manufacturing in general, several Departments of the Government of India and state governments are implementing industrial policy to promote innovation in cutting-edge industries. In the last Union Budget, the Government of India announced a Rs 1 trillion funding pool to spur private sector-driven research and innovation, including a Rs 100 million VC fund to invest in innovation in the Space sector. 

The case for direct public funding of innovation through research and development (R&D) by public institutions is not under dispute. So also is that for tax concessions to private sector investments in R&D. But there’s a challenge with public funding of startups involved in technology innovations. What’s the appropriate mechanism to do such early-stage funding?

For a country like India, there’s a strong economic and strategic imperative to encourage entrepreneurship in emerging technologies like artificial intelligence, quantum computing, cyber security, 5G and 6G telecommunications and its use cases, EV battery chemistry, green hydrogen production, etc. While the country has witnessed a sharp rise in venture capital investments, a very tiny proportion has gone to these cutting-edge technologies. There’s a strong case, therefore, for governments to step in and address the market failure by financing startups involved in the prioritised sectors. Barry Naughton describes here how the Chinese have gone about it. 

Traditionally, public support for startups has been to establish incubators and provide space and shared services. While that remains the situation, a few governments are tentatively venturing into financing startups engaged in innovation. 

In theory, such startup financing can come as grants, or debt, or equity. But debt is not considered appropriate for startups since they are unlikely to be able to generate requisite cashflows in the foreseeable future to meet the loan repayment obligations. And as the article above informs, while grants are simple to implement, they come with administrative delays. In addition, grants have an incentive compatibility problem. 

For this reason, it’s argued that public support to support innovation should be as equity. Since making equity investments requires good due diligence and portfolio management capabilities, it’s also argued that equity investments should be made through arms-length entities with professional investment managers. Besides, equity investments directly by government entities create insurmountable legal issues and are vulnerable to being abused. 

Given all the aforementioned, how do governments support startups in prioritised sectors? 

As a starting point, the funds must be routed through an arms-length entity. Government departments themselves or units housed within departments should not be making grant approvals. The appropriate entity can be a society or a company, depending on the instruments proposed to be deployed. 

The choice of instrument is more complicated. It’s natural to be attracted by the venture capital (VC) model and replicate it with public funds. There are several challenges with this approach. For one, to avoid public oversight agencies (with all the accompanying risk aversion it induces), like the National Infrastructure and Investment Fund (NIIF), the entity must have a majority of private shareholders. 

But, if a NIIF-like entity is merely blending public and private funds instead of offering public funds as a concessional layer, it’s unlikely to serve the purpose. For a start, there’s little by way of de-risking achieved by the public funds deployed. Second, since public funding of startups is necessitated by the market failure in the supply of private capital, this entity is unlikely to make many investments.

It’s for this reason that NIIF has struggled to make any headway in genuinely de-risking infrastructure projects and has instead become just one more infrastructure investor in the market. The de-risking objective of public funds will invariably get squeezed out by the returns maximisation objective of private capital in any entity where private shareholders are a majority at the asset management company level. 

Instead, if public funds are blended as concessional capital through co-financing by a wholly government-owned entity, its administration will be fraught with problems. The nature of the instrument used and the degree of concessionality provided will need to be determined on a case-by-case basis, thereby making it excessively subjective and inviting public scrutiny and associated controversies. This possibility invariably creates risk aversion. 

Further, being early-stage firms, the amounts proposed to be disbursed are small. Since equity investments demand a certain level of legal, governance, and financial due diligence, the due diligence-related transaction costs for smaller equity investments are disproportionately high, even prohibitive. 

The option of having an entity which is fully owned by the government and using public funds to undertake equity investments also cannot avoid the accompanying rigorous public oversight. This creates problems since the entity will be making genuinely risky investments, and the vast majority of them will fail. Hindsight-based scrutiny of such investment decisions can be fraught with recriminations. Traditional public oversight and VC-type investments cannot go together. 

Further, given the constraints of a purely publicly financed entity, it’s tough to recruit and retain the minimum complement of a competent team of investment professionals to do due diligence (supplemented with a panel of external experts) on the investments, approve them, and undertake even light-touch portfolio management activities. 

In these circumstances, the most effective public policy instrument to support innovations would be conditional grants with clawback provisions. As mentioned earlier, grants are easy to target and can be structured with conditions, including recovering them in full or partially if the startup succeeds commercially. The CHIPS Act has provisions that require the sharing of any upside beyond a certain level with the government. Similar upside-sharing provisions exist in all the innovation-focused funding across Europe. Besides, unlike equity or debt, grants are easier to administer for both the grantee and grantor. 

Finally, grants with upside-sharing provisions offer a win-win proposition to both entrepreneurs and governments. It leaves start-ups without any immediate repayment liabilities, especially attractive in the early days of a startup pursuing risky innovations. It offers governments the possibility of being able to recover the grant partially or fully. In other words, grants with clawback are both incentive-compatible and provide the most cost-effective derisking. 

The entity could prioritise its portfolio management role by bringing together panels of investors, industry leaders, and academic researchers to advise and mentor its grantees. It could have partnerships with VC and other investors to connect its grantees with those looking for good investment opportunities. Being government-owned, it could facilitate various government-related approvals and permissions and also help with leveraging any other support from the government for the startups. 

Given the significant amount of fiscal benefit being provided to individuals/firms, grantmaking must have reasonably rigorous due diligence and governance requirements. Also, given the difficulty of attracting high-quality talent, the in-house team should be combined with a standing panel of reputed subject experts drawn from academia, industry, and the investment world to conduct due diligence on grant applications. 

The technology readiness levels (TRL) for which the investments are eligible should be clearly defined, with the corresponding upper limits for the grant amounts. There should be follow-on grant provisions for exceptionally promising innovations but with higher-level approval. A good practical framework to figure out which stage of innovation is appropriate for funding is below.

There should be standardised checklists and templates for the due diligence of the innovation and its technology, the business model, the entrepreneur, the legal issues, etc. Due diligence should involve 2-3 screenings, with the final short-listed applications subjected to reasonably rigorous scrutiny. The processes should be workflow-automated and high standards of governance followed. A high-level committee should periodically review the grant-making process, the pipelines created, the portfolio management activities, and the health of the grantee startups. 

Such grant-making can be supplemented by offering public funds as a concessional layer to mission-aligned private investment funds with certain light-touch conditions on the destination of the funds. This fund-of-funds approach is easier to administer, given the limited number of transactions. Besides, it enables the recipient funds to more effectively de-risk their investments, thereby expanding their universe of investible startups and crowding in more private capital. 

Weekend reading links

1. The advent of container shipping in the 1950s while dramatically improving productivity also led to large-scale job losses.

Before then, the work of loading and unloading a cargo ship was a slow and dangerous process that frequently consumed several days. Dockworkers grappled with three-dimensional jigsaw puzzles while trying not to be crushed to death by shifting freight. They struggled to configure ill-fitting assortments of goods, positioning sides of beef alongside barrels of liquor, drums full of chemicals and bales of cotton. Unions applied their power over the pace of loading and unloading — or to stop the enterprise altogether — to extract hefty wages. The use of containers dramatically simplified the process, while reducing the number of working hands needed to do it. Suddenly, cargo could be loaded at factories into standard-size steel boxes that could be carried by truck and by train to ports, and then hoisted atop vessels by cranes.

On the day in April 1956 when local officials gathered at the Port of Newark to watch the departure of the first container vessel, they celebrated a milestone in American industrial efficiency. But a top official at the International Longshoremen’s Association, the union that represented East Coast dockworkers, looked on in horror... In the decades since container shipping began, union members have been focused on the equation that propels it. When the first ship set sail from Newark to Houston, it cost nearly $6 a ton to load a cargo vessel by hand, according to “The Box,” a book about the history of the shipping container by Marc Levinson. Soon that cost would drop to 16 cents, with most of the savings produced by trimming the need for longshore workers. Over the last quarter-century, a period of extraordinary global economic integration, the number of workers formally employed in handling American marine cargo has grown to about 64,000 from nearly 41,000, an increase of roughly 56 percent, according to Labor Department data.

2. Striking graphic that captures the extent of China's renewable energy investments compared to the rest of the world.

Despite these investments, China's carbon emussions have surged.

In the two decades after the turn of the century China’s emissions from burning of fossil fuels rose about 245 per cent to around 11 gigatonnes by 2021 — more than double that of the US, the world’s second biggest polluter. Among their top worries are the slow rate of retirement of older coal-fired power plants, a resurgence in the pace of new coal builds — China accounted for two-thirds of all global coal-capacity additions last year — and a move last year by Beijing to guarantee a fixed payment to coal power stations, rather than just pay for the energy they produce.

3. How Ozempic and Wegovy are transforming the use of gyms in the US.

Gyms are pushing their stair climbers and fixed weight machines to the periphery and replacing them with open space for body-sculpting classes, free weights and individual training sessions... As the drugs gain acceptance, fewer people are likely to rely on exercise as their primary weight loss tool and the drugs’ side effects, nausea and intestinal distress, can make high-impact cardio activities uncomfortable. However, GLP-1 users still need the gym. Studies suggest that the drugs cause significant muscle loss along with fat, leading to problems with balance and mobility as well as saggy skin sometimes dubbed “Ozempic butt”. Strength training seems to be the answer not just for GLP-1 users but everyone else. A growing body of medical literature suggests strength training cuts mortality, particularly for women, while also helping to prevent osteoporosis and relieving the symptoms of depression.

4. On related note, Eli Lilly is now at the forefront of the race to bring out the first weight-loss pill. It's estimated that there'll be 16 new obesity drugs launched before end of the decade by the likes of AstraZeneca, Pfizer, and Amgen. 

With the most potent weight-loss shot and a pipeline of 11 experimental treatments, including what is widely expected to be the first approved small molecule GLP-1 pill, Eli Lilly stands to be the biggest winner in a market that is projected to grow to $130bn a year in peak sales by the end of the decade... In 2018, after Swiss drugmaker Roche turned down the rights to license a promising GLP-1 pill to treat type 2 diabetes from its sister company Chugai, a rivalry dating back more than a century boiled up once again. Eli Lilly beat out its Danish competitor Novo Nordisk for the rights to the experimental drug after a short bidding war, paying just $50mn upfront, according to two people familiar with discussions. Novo Nordisk declined to comment. Skovronsky could not recall whether the pill’s potential as a weight-loss treatment was even discussed at the time of the licensing deal. But the pill — now known as orforglipron, which looks set to be first small molecule anti-obesity pill if it launches as planned in 2026 — is one of several fronts in which Eli Lilly appears to be outmanoeuvring Novo Nordisk for supremacy in the weight-loss drug market.

This about the competition between Eli Lilly and Novo Nordisk

In 1923, Eli Lilly was first out of the blocks with a commercial insulin product to treat diabetes, which until then was considered a death sentence. Novo, then a standalone company before its merger with Nordisk, created a longer-lasting version and the first insulin pen. In 1982, Eli Lilly launched the first synthetic, mass-producible version of human insulin. In 2005, Eli Lilly then created the first GLP-1 drug — a twice-daily injection, but Novo Nordisk would revolutionise the market with the launch of Ozempic in the US in 2017. Despite Novo Nordisk being first to market, Eli Lilly has benefited from “a second mover advantage” with the launch of its weight-loss medicines, says Rajesh Kumar, head of healthcare equity research at HSBC. “They can see what traps the guy ahead of them is falling into,” he says, allowing them to ramp up manufacturing faster and to invest in next-generation products. This year, Mounjaro and Zepbound, which are both based on the active ingredient tirzepatide, are set to generate $18.8bn in sales between them, according to analyst consensus estimates — edging closer to Novo Nordisk’s $27bn in projected revenues from Ozempic and Wegovy, despite being on sale for a shorter period of time. Sales from Eli Lilly’s GLP-1 franchise are projected to surpass Novo Nordisk’s by 2027.

5. India is set to eclipse the US and become the country with the largest pool of software programmers.

Indian programmers are also the cheapest talent.

India's IT sector (and GCCs) in numbers

The it sector is a juggernaut, generating about $250bn in annual revenues, or 7% of GDP. But GCCs are increasingly important, too. The country now hosts some 1,600 of them. Amazon’s biggest office in the world is in Hyderabad. A fifth of Goldman Sachs’s staff are in India, as are a fifth of the world’s chip designers. New GCCs are opening at a rate of roughly one a week... they are thought to employ some 1.7m of its IT sector’s 5.4m workers, with salaries over four times the national average. By one estimate, they create about $120bn in value and are growing by 11-12% a year. If so, GCCs already represent over a third of India’s services exports, which would make them its biggest export category after it services themselves.

The first article also highlights that software programming is the one sector where generative AI appears to be delivering on its promise. 

One reason why generative AI can be especially useful for developers is the availability of data. Online forums, such as Stack Overflow, hold enormous archives of questions asked and answered by coders. The answers are often rated, which helps AI models learn what is helpful and what is not. Coding is also full of feedback loops and tests that check if software works properly, notes Nathan Benaich, of Air Street Capital, a venture-capital (VC) firm. aimodels can use this feedback to learn and improve. The consequence has been an explosion of new tools to help programmers. PitchBook, a data provider, tracks some 250 startups making them. Big tech is leading the charge. In June 2022 GitHub, which is owned by Microsoft, launched Copilot. Like many tools it can, when prompted, spit out lines of code. Around 2m people pay for a subscription, including employees at 90% of Fortune 100 firms. In 2023 Alphabet (Google’s parent company) and Meta (Facebook’s parent) released rivals. This year Amazon and Apple followed suit. Many firms have built ai coding tools for internal use, too.

6. Excellent Bloomberg long read on how China went past the US to become the solar cell manufacturing monopolist.

7. More disappointing experience from the role of private equity investments in retail. Asda, the UK's third largest supermarket which will turn sixty next May, was taken over in 2020 through a leveraged buyout by the PE firm TDR Capital for £6.8bn. It's now struggling to survive on the face of competition from Aldi and Lidl. This is only the latest example of problems with PE investments in retail, raising questions about PE in retail sector. 

It remains fair to ask whether private equity ownership of supermarket chains is a good idea. Clayton, Dubilier & Rice acquired Morrisons for nearly £10bn in a deal later called a “fiasco” for the banks involved. Morrisons made a £1bn loss last year because of its high debt costs and Rami Baitiéh, former chief executive of Carrefour France, is now trying to revive it. Supermarkets are fine targets in theory, with famous brands and strong cash flows. But competition is too intense to deploy the old private equity strategy of cutting costs and raising margins: shoppers can too easily go down the road (or online) for better bargains. The only way to succeed is by investing enough to increase sales and seize market share.

Also on excessive leverage assumed by PE firms

As private equity groups struggled to exit investments, they had been using more complex and opaque forms of borrowing, such as leveraging up funds that own several already-indebted companies to finance payouts for their investors. “Recently as exits have been difficult, I think [buyout groups] needed to go back to financial leverage” to make money, David Hunt, the chief executive of $1.3tn asset manager PGIM, told the Financial Times. “That can help you on the upside” but it could “really accelerate things on the downside”. “It’s complicated enough that many people don’t understand it,” said Hunt, adding regulators should insist on more disclosure about such debt. “I think [introducing] some common way of understanding how much leverage is in the system is a good idea.” Buyout groups have for decades loaded debt on to the balance sheets of the companies they buy in order to pay for their acquisition. In recent years they have been increasingly using so-called net asset value loans, where a buyout fund borrows against the stakes it holds in those companies. The entities that manage those funds can also take on debt, representing another layer of leverage. Groups including Vista Equity Partners, Carlyle and Hg Capital have used NAV loans to pay dividends. The loans can also be used to support struggling companies within a fund. However, they are controversial because they add more leverage and cross-collateralise the fund’s investments, meaning that troubles with one company can spill over to others and put the entire fund’s returns at risk.