• Understanding overlooked social factors, from local community relations to AI risks, is relevant to long-term investment outcomes.

• Simple, metrics-driven ESG frameworks often miss deeper social drivers that ultimately shape company performance.

• The interplay between community buy-in, regulatory shifts, and geopolitical realities reveals hidden risks and undervalued opportunities.

  
  

John Donne

It has now been almost three years since we refreshed and expanded our Active Ownership Report to include regular qualitative discussion of some key issues affecting both our portfolio and wider industry debate. The eleven quarterly reports we have produced since then have covered topics ranging from the geopolitics of the energy transition, the war in Ukraine, analyses of industries including wind, solar and shipping, country risk in China, and several articles on the evolving responsible investing (‘RI’) paradigm. Within the report, we have also discussed governance issues such as incentivisation.

We have focused on these issues primarily because we feel that they have the most material relevance to our portfolio. In our view, for non-impact managers, responsible investment (RI) is the process of incorporating the consideration of long-term, often non-directly financial factors into our analyses and engagements. This is designed to support the generation of sustainable, long-term returns for our clients. However, as a public-facing document, it is also the case that the issues we focus on in this report are those about which we receive the most questions, and which are most prominent in broader industry conversations.

As such, environmental and governance-related issues have dominated. This is not terribly surprising, as the ‘E’ and ‘G’ in ‘ESG’ have been at the centre of responsible investment analysis and reporting over the past several years. Media-friendly, (until recently) generally uncontroversial, and importantly easy-to-quantify, these themes have dominated RI discussions. Meanwhile, social (‘S’) issues, which are often considered hard to measure and define, have hovered on the sidelines, overlooked and unloved.

‘ESG’ has always been a group of awkward bedfellows: some things which should be there are missing (geopolitics), others which are there arguably don’t belong (governance), and in the middle of it all is a separate category for ‘social’, even though we would submit this describes the entire entity. After all, it is the social impacts of things like governance and environmental policy which ultimately impact value creation. This piece will briefly examine why ‘S’ has lagged in responsible investing frameworks before highlighting several emerging social issues, including ‘licence-to-operate’ community relations and artificial intelligence (AI) safety, which we believe will move increasingly to the forefront of the debate in coming months and years.

‘S’ metrics: Path of least resistance?

Unlike measuring a carbon footprint or scrutinising executive pay structures, social issues are often inherently multifaceted. They span a vast array of stakeholders – employees, communities, consumers, and supply chains – making them notoriously difficult to define and quantify. The absence of standardised reporting and readily available benchmarks has meant ‘the S factor’ has lacked a clear link to corporate performance. This is especially the case when a range of individual issues are aggregated into a single ‘score.’

The incentive to simplify and quantify all elements of ESG has had a direct effect on the types of social issue that have become the centre of attention. Diversity, equity, and inclusion (DEI) statistics, for example, now routinely appear in annual reports. Proxy agencies have been quick to apply board and management social diversity quotas, to which their powerful voting recommendations defer. Elsewhere, statistics covering CEO-to-worker pay ratios, workplace injury rates, and community-related philanthropy (often literally provided as a dollar value) have become mainstays in corporate sustainability reports. These measures emerged in response to real problems. Often, they address an entrenched inequality and have almost certainly helped to catalyse at least directional reform (see Figure 1). Many boards are more diverse now than they were a decade ago, and, in select cases, pay disparities have narrowed.

At Hosking Partners we recognise the benefits of enhancing diversity and properly aligning management incentives, but in practical terms, rigid, one-size-fits-all approaches – whether by imposing strict board diversity quotas or reflexively voting down executive remuneration schemes – can lead to perverse outcomes. For instance, rejecting a well-conceived pay plan that incentivises strategic risk-taking at precisely the moment a company may need it most could harm returns. We recently wrote in more detail about the importance of long-term incentive metrics that encourage management to act like owners. Equally, moving to deselect an outstanding board member because of insufficient diversity metrics, without considering that director’s unique contribution, may deprive the firm of vital expertise.

As ever, the devil is in the detail, and any responsible manager has a duty to their clients to understand and respond to that complexity. We therefore approach these issues on a case-by-case basis. While our position often aligns with proxy agencies on questions like board diversity or excessive pay, we reserve the right to deviate if the nuance of a specific situation suggests otherwise. This allows us to balance the financial interests of the company against longer-term, intangible factors. The recent backlash against some elements of ‘ESG 1.0’ is at least in part a result of the way these issues have been oversimplified and applied in a standardised way that is divorced from case-by-case due diligence.

Meanwhile, amidst the focus on a narrow set of easily quantified issues, other social themes have remained comparatively neglected. Against this backdrop, the next few sections will explore how emerging challenges are set to reshape how we approach the ‘S’ in ESG. In particular, we will highlight the social flipside of decarbonisation, and examine how Big Tech’s sustainable credentials, often magnified by careful curation of simplistic and quantifiable metrics, are moving ever faster into the spotlight.

Energy transition: A social affair?

Writ-large, decarbonisation is a global journey. We have previously noted that the molecular contents of Earth’s atmosphere pay little attention to international borders or individual government policies. But while the overall challenge is global and top-down, the distribution of its effects is disparate, uneven, and bottom up. For this reason, the energy transition remains a local, and therefore intensely social, affair.

Nowhere is this more evident than in mining and natural resource extraction. These are sectors whose ‘licence to operate’ hinges on maintaining goodwill within nearby communities. Whereas broader responsible investing conversations often pivot around global policy goals or macro-level emissions statistics, these do not always capture the on-the-ground realities: indigenous land rights, environmental justice protests, and local anxieties about job security can derail even the most well-funded ventures. If a community feels overlooked or exploited, its resistance can be the difference between a project’s success or a costly stalemate (the ongoing Cobra Panama mine debacle is an example of such a dynamic).

Reassessing decarbonisation ambitions through this local lens helps illuminate the trade-offs inherent in the clean-energy transition. Solar panels, batteries, and EVs are crucial to weaning economies gradually off hydrocarbons, yet these require critical minerals – copper, nickel, lithium, cobalt, PGMs – whose extraction is not only energy-intensive but fraught with social tensions. The push to secure these resources often sees multinational mining firms operating in regions far from their shareholder bases, leading to a mismatch between local and global priorities. Striking the right balance between shareholder returns and community well-being is no small feat, particularly in jurisdictions with weaker governance frameworks or complex socio-political histories.

This tension between local interests and global decarbonisation targets only looks set to intensify. Conventional energy sources are still needed as we move towards a cleaner future: many renewable technologies remain contingent on consistent baseload power, not to mention the heavy equipment and logistics essential for building out energy infrastructure. These ‘bottom-up’ realities are shaping the form the top-down energy transition is taking. Five years ago, experts on both sides of the debate would have balked at the idea that both coal and solar would surprise to the upside out to 2030, but this is what we are seeing unfold (see Figure 2). Such results seem shocking compared to ‘the narrative’, but inevitable once you consider the real-world interaction of economics, technology, geopolitics, and socio-political dynamics.

This latter factor is both particularly important and considerably under-analysed. Without popular support well-intentioned reforms can quickly become politically untenable, especially in lower-income communities. In the context of energy, this has led to a backlash against renewables as an oversimplistic narrative about clean energy being intrinsically cheap (even “free”) buckled under the real-world cost pressures of an accelerated transition and poor implementation. The growing popularity of transition-sceptic governments across the West is partly a result of this simple economic impulse.

The effects of this sort of divergence between what people are told to expect and what they actually experience are magnified by social media. The algorithms which underpin this mode of communication are incentivised by engagement – which drives advertising revenue – and countless studies show that engagement rises with emotions like disgust and outrage. (1) Polarisation seems, at least in the short term, good for Big Tech. Unfortunately, it is terrible for achieving the sort of cross-society consensus required to execute long-term transformations like an energy transition.

All this points to a shift in how businesses manage the energy transition, with a greater emphasis placed on adaptation and resilience. Given the possibility that global emissions targets will be missed or delayed, there is growing recognition that supply chains need to harden critical infrastructure against climate volatility. Importantly, this adaptation imperative applies not only to physical capital (factories, power infrastructure, etc) but also to the human capital upon which companies rely. Energy transition is a primarily technological feat – substituting one power source for another without sacrificing efficiency – but climate adaptation is first and foremost a social endeavour. It requires buy-in from local stakeholders, robust consultation, and equitable distribution of benefits. This will become a key responsible investment theme in coming years.

In our own portfolio, we see these trends play out across multiple industries. In mining, for instance, operators such as Sibanye-Stillwater in South Africa have discovered that the right to develop local mineral resources is contingent on forging strong relationships with local communities—a dynamic that will likely intensify as demand for critical minerals grows. This is something that we experienced first-hand during a research trip late last year and Django Davidson wrote about in a recent Hosking Post examining the capital cycle in PGMs. The same logic applies to energy, metals, and shipping: social licence is not a peripheral concern but a core driver of operational stability and, by extension, long-term shareholder returns. This is an area of focus for our ongoing engagements, and one to which we will return in more depth in coming quarters.

Big Tech: Golden age or regime change?

Many of today’s tech giants boast surprisingly high marks on standard ‘ESG ratings’, in part because their carbon footprints appear relatively modest compared to heavy industry, and their governance structures seem robust. Yet this picture belies a host of social controversies which ratings agencies struggle to measure and weigh. These include labour disputes in gig economies, concerns over data privacy and misinformation, alleged monopolistic practices, rapidly growing evidence concerning the negative effects of social media algorithms which profit from polarisation, and growing unease over the ethical and safe deployment of ever-more-sophisticated artificial intelligence (AI) systems.

AI safety, in particular, is fast emerging as a critical priority. Advanced machine-learning models increasingly shape everything from online advertising to public discourse, healthcare diagnostics, judicial decisions, military targeting, insurance premiums, policy recommendations, and more. However, while these algorithms promise greater efficiency, clarity, and speed, they also introduce risk. Even discounting Hollywood-esque (but nonetheless real) concerns over bioweaponry, autonomous robotics, and cyberwarfare, AI tools will also have far more subtle effects which could embed biases, undermine individual autonomy, amplify disinformation, disrupt the job market, and even prompt fundamental changes in how the human brain works. It has been shown that relying on GPS for navigation reduces naturally occurring spatial memory in the hippocampus, (2) while heavy use of social media affects everything from our capacity for language development to the reward pathways which govern addictive tendencies (3). It seems likely, therefore, that increasingly outsourcing a vast range of cognitive tasks to AI will prompt further alterations (for better or worse). This is a rapidly developing field, and one which conventional responsible investing frameworks have thus far failed to integrate into their often simplistic, metrics-driven models.

Why should investors care? There are two angles to this, one which is primarily financial, and the other regulatory. We illustrate both below.

During the lead-up to the second Trump presidency we have witnessed a remarkable ‘rush to Washington’ by the pre-eminent US tech CEOs. Millions of dollars in donations, conciliatory policy about-faces on issues like content moderation, and – perhaps most interestingly – an uptick in personal physical presence. From Mar-a-Lago to Washington’s St John’s Chapel, where the inauguration church service took place, it has become commonplace to spot the faces of not only Elon Musk, but Bezos, Zuckerberg, Sundar Pichai (et al) lurking in the background, eager to signal their (often apparently newfound) allegiance to the incoming administration.

What exactly is going on here? Media commentators have been quick to point to a new “golden age” of technological oligopoly, heralded by the arrival of AI and Trump’s supposed willingness to deregulate. But something about this narrative seems off. Firstly, if both observations are true, and the incumbent tech CEOs foresee another decade of market dominance, easy cashflow, supernormal margins, and hand-off regulation, then why rush to Washington at all? What’s the point, if everything is coming up smelling of roses anyway? Secondly, has Trump really demonstrated a hands-off approach to Big Tech? The jury is out. As ever, it is difficult to separate his rhetoric from actions, but Trump’s first term certainly prompted some of the most significant crackdowns on Big Tech power of the past two decades. In addition to frequent, public criticism of every company from Amazon (tax avoidance), to Google and Meta (search and content bias), Trump 1.0 also oversaw the most significant Department of Justice probe into Big Tech’s monopolistic practices since the 1990s, as well as an executive order questioning certain liability protections granted to tech companies by Section 230 of the Communications Decency Act. Notably, Trump has indicated he would like to readdress and expand both matters in his second term.

Furthermore, the rapid iteration of AI is already catalysing an additional layer of regulatory attention. A large part of the way Big Tech has maintained market dominance has been by selling targeted ads or facilitating their sale. Remarkably, since 2008, Big Tech’s share of total US advertising revenue has doubled to over 65%. (4) It benefits the companies to give access to many of their services for free (or close to), to increase the size of the engagement audience. This makes sense because often the basic services being offered (e.g. search, photo sharing) are only incrementally useful in and of themselves, and there are many competitors. Their elasticity of demand is high. If you introduce an access cost, you can only raise it so high before customers switch service. As such, most people tolerate ads in return for near-free access to the service itself. AI-driven search and interaction – voice assistants, generative AI, recommendation algorithms – may change this ad ecosystem entirely. This could take several different shapes. On the one hand, we could see a shift from ‘search results with ads’ to ‘answer engines’ that have fewer or differently placed monetisation options. This is because of the sheer utility of AI tools, where elasticity of demand is lower (demonstrably, OpenAI is reportedly considering a $2,000/month access fee for its newest GPT model, although it remains to be seen whether such ideas survive contact with open-source models like DeepSeek, discussed more below). Such ‘walled garden’ approaches will inevitably invite scrutiny regarding the socio-economic stratification of access to AI. On the other hand, even where less powerful, ad-assisted, free-to-use products continue to be sold, we are likely to see a wave of regulation aimed at limiting how Big Tech uses AI to increase the depth and reach of its advertising. Such shifts cannot be fully understood solely by tweaking inputs to financial models. The social impacts of these changes also need to be carefully and qualitatively evaluated.

A second emerging issue in AI is rooted in the logic of the capital cycle. This has been thrown into sharp relief recently by the release of China’s DeepSeek model. The AI revolution is prompting an innovation upcycle of enormous proportions, with capital flooding into the sector. This is leading to an influx of competitors, and in turn forcing the incumbents to ramp up their own capital expenditure to attempt to maintain market share (see Figure 4). All else equal, the capital cycle approach tells us that this should lead to downward pressure on the average returns this capital generates. Demonstrably, in 2024, we estimate the Big Tech firms have around $600bn of AI-related invested capital chasing an LLM market currently worth just $6bn. Even assuming highly aggressive revenue and margin CAGRs, the return on that capital may not reach 10% until well into the 2030s. (5) While it is true that strong margins and balance sheets lend these companies resilience, their remarkable market concentration – just five firms account for 25% of all US equity value (6) – increases their sensitivity to small shifts in the assumptions underlying those valuations. When a company is priced to perfection, the marginal effects of underperforming expectations can prove non-linear, as witnessed when $600bn – about equal to the GDP of Sweden – was wiped off NVIDIA’s market cap in a single session following media speculation on the implications of DeepSeek’s claims about how its model was trained.

There is much more to be said about DeepSeek, but for the purposes of this article it serves the simple purpose of demonstrating how fundamentally socio-political issues – such as the implications of open-sourcing, or US-China geopolitics – can amplify financial considerations to spread ripples through markets. This is especially the case when they relate to themes that loom large in both the public consciousness and purse. In the retrospective context of DeepSeek’s announcement, and its implications for the vast sums of capex already committed by Big Tech, perhaps the CEOs’ ‘rush to Washington’ can be seen in a different light. Could it be that this behaviour is not bravado, but rather rearguard defensive action? The disturbance caused by this AI-driven capital cycle is already prompting the emergence of a new generation of competitors, and history tells us that generational disruption in a concentrated market tends to prompt a period of market broadening. In such a market, Hosking Partners’ contrarian, diversified and differentiated global strategy seems well-positioned versus historically concentrated indexes.

Conclusion

Our focus on the ‘forgotten S’ within responsible investing goes hand in hand with our capital cycle approach. By looking at social issues through the same contrarian lens that we apply to industries and companies, we aim to identify where sentiment and regulation may be poised to shift, and where undervalued opportunities or unrecognised risks lie. This methodology insists on embracing complexity: rather than relying solely on simplistic metrics, we consider local contexts, supply chain complexities, and the longer-term societal impacts of corporate activity.

Such an approach helps us see the wood for the trees. Where many observers get caught up in short-term headlines or uniform scoring frameworks, we dig deeper to spot the patterns that truly drive long-term value creation. By blending capital cycle principles with a thorough consideration of social factors, we believe we can more accurately gauge both the potential upside and the real-world viability of a business. In this context, we believe issues such as those raised in this article will be of growing importance to investors in coming years, and we look forward to returning to discuss them in more depth in future reports.