Performance review: Narrative versus fundamentals

Markets often seize on a dominant theme and apply it broadly, rewarding what fits the narrative and punishing what does not.

Over the last year, the narrative in emerging markets has been shaped by what some have labelled the “HALO” trade (or heavy assets, low obsolescence). Higher interest rates, geopolitical instability and, in particular, the artificial intelligence (AI) data-centre build-out have fed investor appetite for asset-heavy, cyclical businesses, such as Korean industrials and Chinese state-owned enterprises (SOEs) – areas we tend to avoid on quality and governance grounds. Meanwhile, rapid advances in “agentic”, or autonomous, AI technology have sparked fears of disruption among the kind of asset-light digital companies we prefer.

This shift in market leadership goes some way towards explaining the recent lag in the relative performance of the strategy. Over the 12 months to end-March, the benchmark returned 30.3%, versus 11.6% for the portfolio in US dollar terms.¹ Our analysis shows the index return has been driven by a strong cyclical upswing in earnings across sectors such as metals and mining and technology hardware, combined with significant re-rating of valuations. Re-rating accounted for 12.1% of the index return over the period, compared with 2% for the portfolio.²

The strategy’s returns continue to derive primarily from earnings-per-share growth and dividends, consistent with its outcomes over the longer term (see Figure 1). This matters, because while valuation re-rating is a one-time gain, compounding earnings builds value over more sustained periods. We would expect the strategy’s performance to turn as cyclical tailwinds fade and investors return their attention to the durability and repeatability of earnings and cash generation.

In the meantime, rather than chasing parts of the market where popular stocks are being bid upwards, we continue to focus on fundamentals. We are adding to those we view as high‑quality holdings where prices have fallen faster than intrinsic value, and trimming where cyclical momentum and re‑rating have pushed valuations higher than we think the earnings and growth outlook justify.  

In this review, we highlight the key drivers of recent returns, both positive and negative; explain our current positioning; and outline why we remain confident in the portfolio’s longer-term prospects despite near-term volatility. In a forthcoming letter, we will explore the market’s concerns over AI disruption and the impact on our holdings in more depth.

Figure 1. Drivers of strategy returns (top) vs the index (%), 2018-2025

FSSA GEM Focus strategy

MSCI Emerging Markets Index

Source: FSSA, FactSet, MSCI, March 2026. Note: EPS = earnings per share (net profit attributable to shareholders divided by the number of shares outstanding). NTM = next twelve months, based on analyst consensus estimates. Re-rating refers to the market assigning a higher or lower valuation multiple (in this case, the forward-looking price-to-earnings ratio) to a company, reflecting changes in investor expectations or market conditions rather than changes in earnings alone.

Starting with positive contributors to performance, we saw solid outcomes across very different parts of the market. Chinese hotel operator H World and Peruvian bank Credicorp might seem to have little in common on the surface, but they share some key characteristics: both are beneficiaries of expanding domestic consumption and increasing technological sophistication in their respective customer bases.

H World has a predominantly asset-light franchising model, under which franchisees are responsible for the bulk of capital investment in real estate, and pay recurring fees for H World’s branding and distribution know-how (often via “manachised” arrangements where H World also provides operational support via an on-site hotel manager). This has enabled it to scale quickly and take advantage of robust demand for travel – one of the few consumer categories in China to have held up well during the demand slump of recent years. 

Part of H World’s success lies in its digital offering. Technology has been deployed throughout the business, from the self-service kiosks in its hotels to its app-based membership scheme, and this combination of digital capabilities and scale gives H World a compelling advantage over its rivals in China’s fragmented hotel industry. Our research shows that hotels under the H World brand tend to earn higher revenue per available room (RevPAR) than independently operated ones – a compelling reason for hotel owners to sign up. 

The company has admitted missteps in recent years, including a poorly timed acquisition of Germany’s Deutsche Hospitality in 2020; H World is also trying to divest from some legacy leased and owned properties to become more of a pure asset-light operator. RevPAR has now turned positive after a downturn, and the underlying metrics are good, with H World posting robust free cash flow and a return on equity (ROE) of more than 40% for 2025.³ However, the stock now looks a little expensive after its strong recent run, and we have been trimming our position. 

Credicorp has also risen over recent months after posting strong results that showed healthy net income growth and a meaningful improvement in asset quality, thanks to a decline in the non-performing loan (NPL) ratio. Like H World, Credicorp’s embrace of technology means it has key competitive advantages over its peers. We have long been struck by the company’s willingness to “self-disrupt” by investing in innovations that can open up new markets and deepen customer relationships, rather than simply defending the existing business. 

That cultural approach has been a feature for years, but it is becoming increasingly evident in results, particularly through Yape, the group’s payments and services platform. Around 15 million users (almost half Peru’s population) use Yape, and it also has a market-leading share in person-to-person payments. Related initiatives include Credicorp’s subsidiary Tenpo, which recently became the country’s first licensed digital bank. 

Our investment in Credicorp illustrates our preferred way of gaining access to commodity-driven economies such as Peru. In many cases, the energy and materials companies that dominate these markets come with governance, state ownership or capital-allocation risks that we would rather avoid. So instead, we often look for high-quality beneficiaries in the same economy – such as dominant banks or consumer franchises – where we can still participate in improved confidence and activity, but through a more resilient business model. Credicorp, with its strong position in its domestic market, fits that profile: it has a roughly 30-40% share of deposits and assets in Peru.

The build-out of data-centre capacity and rapid development of AI capabilities have influenced performance in both directions over the last year.  

Starting with the perceived winners, the beneficiaries of massive hyper-scaler capital expenditure include the major Taiwanese and Korean chipmakers, whose products are used to train advanced AI models. Among them are our holdings Taiwan Semiconductor Manufacturing (TSMC), Samsung Electronics and SK Hynix. These firms have seen their revenues and share prices surge, and all three have contributed positively to the strategy’s absolute performance over the last 12 months. However, we discern important differences in their business models and competitive moats, and this is reflected in the composition of the portfolio. 

TSMC’s unrivalled position at the most advanced process nodes, combined with its deep and longstanding customer relationships, confers clear structural advantages. We see the company as a natural beneficiary of “productisation” – the diffusion of AI into end-devices and industrial applications – which we believe will accelerate over the coming years, even if near-term data-centre spending normalises. Our stake in TSMC remains the largest in the portfolio. 

As for the Korean firms, while both are well-run businesses, SK Hynix remains the leader in high-bandwidth memory (HBM), a type of advanced chip used for AI applications, where it benefits from a first-mover advantage, strong customer base and proven execution. The recent turnaround in Samsung’s business has been impressive, and it has made rapid progress in HBM. But a greater proportion of its revenues are tied to legacy memory chips such as dynamic random access memory (DRAM), a more competitive and commoditised segment where earnings are more cyclical in nature. 

Both SK Hynix and Samsung have reallocated production capacity towards HBM – while this is strategically the right move, it has affected supply for legacy memory chips and consequently led to a spike in pricing and a cyclical increase in these companies’ revenues. This concerns us, as a softening in memory demand could just as easily compress earnings. We have therefore been trimming our exposure to both Samsung and SK Hynix, while continuing to prefer the latter thanks to its advantages in the more-differentiated HBM segment. 

On the other side of the ledger, concerns over AI disruption risk have impacted several of our holdings. These fears accelerated in the first quarter of 2026 after the open-source agentic platform OpenClaw gained traction, followed by the launch of US firm Anthropic’s agentic workplace tools. The consequent de-rating of companies including Chinese internet giant Tencent (and its major shareholder Prosus), online travel agency (OTA) Trip.com, trucking logistics platform Full Truck Alliance (FTA), and e-commerce firms Sea and MercadoLibre (MELI), weighed on performance. In our view, the market reaction has been too blunt; while certain business models may well be exposed, others will stay resilient and perhaps even be strengthened in the AI era.

Tencent is a case in point. In 2025 and early 2026, the company was punished for spending too little on AI. Then, in March, its announcement that it would increase AI spending was coolly received by the market, failing to offset the growing fears over the agentic threat.

But the disruption case seems to be missing a salient fact. Rather than a thin interface layer that an external agent can easily replicate, Tencent’s competitive advantage lies in its ecosystem – WeChat’s social media platforms, integrated payments, content and gaming – where users already complete actions end-to-end, and where it is not obvious that an external agent would improve the experience. In recent meetings with management, we came away comfortable with Tencent’s sensible and measured approach: it is upgrading its research team to try to build a competitive large language model (LLM), but it is also deploying AI to monetise its products in the here-and-now. For instance, AI is improving the company’s advertising targeting and efficiency, and accelerating production and engagement in gaming.

Similar dynamics apply across other holdings that have been hit by perceived AI disruption of late. For example, Trip.com has hard-to-replicate advantages in the form of its access to live inventory, merchant density, room availability and pricing. Customers use its platform not only to book flights and hotels, but also to research the available options and decide from there. Removing the user would also remove the “discovery” part of the trip-booking journey, which is being continually refined as new information comes to light. Additionally, Trip.com offers fulfilment, integration with payments and a track record of service that has enabled it to build trust. These elements are very difficult to replace with an AI solution, which in any case would likely not be able to provide a lower-friction experience to the user relative to the finely tuned experience already on offer. Reflecting this, existing agentic travel agents have proved notably unreliable.⁴ 

In e-commerce, too, the market is treating digital platforms as though they are merely checkout interfaces that can be displaced by an AI agent, whereas in reality companies such as MELI and Sea operate integrated ecosystems combining discovery, logistics, payments, credit, advertising and merchant tools – and will therefore be much more difficult to disintermediate. While both companies have experienced margin pressure due to their continued investment in their businesses, we think the market is underestimating the extent to which this spending is reinforcing their core strengths and bolstering their ecosystems – thereby improving their resilience to both AI disruption and competition from e-commerce rivals.

With these factors in mind, we see the recent de-rating of internet companies as more about valuation compression driven by AI fears than evidence of lasting deterioration in the underlying businesses. Our conviction is strengthened by our analysis that shows earnings remain robust among these companies (see Figure 2). The disconnect between the market narrative and the underlying reality has created an attractive opportunity, and we have been adding on weakness to our positions in these names.

Figure 2. Robust earnings among internet platform holdings

Source: Bloomberg, April 2026.

Beyond the internet companies discussed above, the most significant detractor from recent performance was HDFC Bank, which was weighed down by the impact of the Iran war and related disruptions to energy supply (India is a net importer of oil). 

HDFC Bank further weakened in mid-March following the resignation of the non-executive chairman. Our subsequent research into the circumstances has been reassuring: the episode appears to have reflected friction over the boundaries of a non-executive role, rather than the surfacing of deeper governance or operational issues. This view was reinforced by the Reserve Bank of India’s unusual public statement that it had found no evidence of ethical concerns implied by the chairman’s letter. HDFC Bank’s shares have recovered well since, and recent financial results and our engagement with management also suggest no impact on asset quality or growth from the war. 

Taking a broader view, India brings together many of the themes that have shaped markets recently, from the threat of AI disruption in its IT services industry to the energy price shock. We continue to view India as one of the most fertile hunting grounds for high-quality companies in emerging markets, and the energy, talent and ambition we see on the ground are hard to match. 

That said, our conclusion after our most recent trip, in February 2026, was largely unchanged. While valuations have come down, the companies we like are still not cheap, and the market continues to price a relatively strong growth trajectory even as the macro backdrop looks more mixed. Over the long term we remain confident in India’s prospects, but near term we still see challenges. The key macro risk is that the IT-services engine for white-collar job creation and urban consumption is slowing, at the same time as parts of consumption appear increasingly supported by credit rather than income. That mix could translate into weaker discretionary demand, softer housing demand in key IT corridors and reduced support for the rupee. Politically, reform momentum has also eased; while a major shift in policy is not our base case, it is an underpriced risk. 

A useful way to frame valuations is market capitalisation relative to GDP: India is currently at around 117%, placing it in the 81st percentile of the past 15 years. While that’s below the peak of 150% reached in late 2024, the most compelling entry points historically have arrived when that ratio has fallen below 70%, and we are still some way from that level.⁵ 

With valuations still elevated, we continue to limit our exposure to India beyond key financial holdings such as HDFC Bank, ICICI Lombard General Insurance and Kotak Mahindra Bank. Our trip did, however, help sharpen our focus on a small number of high‑quality bottom-up opportunities across IT, financial services and consumer platforms, where we are carrying out further research. 

Looking ahead, the big question in the short term for India – and for emerging markets more broadly – is the outcome of the Middle East conflict. Around 20% of global seaborne oil and liquefied natural gas (LNG) flows through the Strait of Hormuz, so the current disruption has immediate implications for energy prices, freight costs and inflation. This is a particularly significant issue for economies in South and Southeast Asia that rely on oil and gas imports. 

Nevertheless, we expect our portfolio holdings to stay resilient. The companies in our portfolio benefit from strong competitive positions – network effects, trusted brands, deep distribution, cost leadership and high switching costs for essential products and services. These characteristics have historically translated into pricing power and the ability to protect margins even in more inflationary or volatile conditions. They are also, as we have illustrated in our discussion of AI, the characteristics that may determine which firms get disrupted and which do the disrupting. 

On balance, we continue to believe the portfolio is attractively valued relative to its growth potential. Our analysis suggests earnings can compound at a mid- to high-teens rate over the medium term. If the conflict continues, there is likely some downside to this, but we would still expect the portfolio to hold up better than the broader market. The portfolio trades at around 18x price-to-free cash flow versus 22x for the market – a valuation that we view as reasonable for the quality of businesses we own. 

Near-term volatility may persist. But for investors willing to look past the prevailing narrative, we think the opportunity set remains compelling.  

References

¹ Performance shown net of fees. Source: FSSA Investment Managers, MSCI, as of March 31, 2026.

² In USD terms. Source: FSSA, MSCI, as of March 31, 2026.

³ FactSet, as of December 31, 2025

⁴ Research from computer scientists at Princeton University on AI reliability found AI agents were particularly unreliable when used to book travel, with agentic assistants frequently booking the wrong flights and failing to process customer refunds. See “How (un)reliable are AI agents?”, Financial Times, April 2026; and “Towards a science of AI agent reliability”, arXiv, February 2026.

⁵ Bloomberg, March 2026.