Over the last two years we have used the Active Ownership Report to explore some of the debates implicit within the energy transition. For us as long-term investors, understanding conflicting views is important, not only to help position the portfolio to outperform as the transition progresses, but also to help inform our engagement efforts with the companies we own. A common theme has been the complex relationship between the set of risks covered by the ‘E’ in ESG – the environment – and those covered by ‘S’ – social issues. In this context, we have discussed how developed world approaches to decarbonisation could impact the emerging world (‘A diverse world’), how the route to net zero cannot be extricated from the broader role of energy in society (‘The Maze to Net Zero’), and how recent Russian geopolitics may have been influenced by the energy transition (‘The Gambler’). These are big ideas, spanning sectors and regions through time.
In this report, we turn our attention to a more specific example of this E-S tension – the Chinese solar industry. The solar industry is, at its heart, a mining industry. The basic building block of a solar panel is quartz sand, which is dug out of the ground and smelted into silicon metal, purified into polysilicon, upgraded to mono-crystalline polysilicon, and finally upgraded again to photo-voltaic (‘PV’) silicon. Around 10kg of quartz sand produces 1kg of PV silicon. The PV silicon is then processed into wafers, the wafers aggregated into solar cells, the cells into modules, and the modules connected to a variety of ancillary devices such as inverters to produce the final solar panel product. China currently dominates this value chain. If you have solar panels on the roof of your house, the overwhelming likelihood is that at least a part of them was manufactured in China, which produces 90% of the world’s polysilicon, 96% of its wafers, 83% of its cells and 75% of its modules (1). There are several reasons for this dominance, but the underlying driver is government subsidies, which come in several forms.
The most obvious of these is subsidised energy. About 80% of the cash cost of PV silicon is energy. The cheapest energy in the world is subsidised Chinese coal power, at around 2-3¢/kWh, almost half the price of the next cheapest source, domestic US gas, and about five times cheaper than long-run European power prices. This application of cheap coal power to polysilicon production allowed China to build out manufacturing scale, monopolise the lower end of the solar cost curve and rapidly gain market share (see Figure 1). In 2012, about half of the cost of a solar project was manufacturing compared to just one fifth today. This manufacturing scale, facilitated by subsidised coal power, has been the largest driver of the remarkable deflation in solar costs witnessed in the past decade, and have allowed China to price PV silicon at about $8-10/kg. Bottom-up analysis suggests that even with subsidies, a minimum price of $12.50/kg is required to generate a narrow 5% IRR (2). At such prices it is hard to see how producers are earning an economic return, and one study has estimated that without any subsides at all Chinese PV silicon could cost as much as $70/kg (3). An even more controversial source of subsidy than coal power, however, is the alleged use of forced labour – specifically from Xinjiang’s Uyghur population – in the solar value chain.
Of the 90% of the world’s solar-grade polysilicon that is manufactured in China, over half originates in Xinjiang province. This vast region in Western China is home to the majority Muslim Uyghur ethnic group who, over past years, have been subject to a state-sponsored “labour transfer” programme. The Chinese government claims these initiatives are voluntary, but evidence suggests such transfers are carried out coercively, with up to 2.6 million transferees unable to refuse or walk away once relocated (4). Such activity is defined as forced labour by the UN’s International Labour Organisation, as it constitutes "work or service which is exacted from any person under the threat of a penalty and for which the person has not offered himself or herself voluntarily" (5). Since these programmes came to light, much work has been done to examine which goods and products they affect, and which companies may be complicit. While the cotton and tomato picking industries are likely the largest recipients of “labour transfers”, serious allegations have also been made about the solar industry (6). Upstream quartz-mining and silicon smelting are the most clearly implicated due to their reliance on manual and low-skilled labour. However, due to the highly integrated nature of the industry, it is difficult to establish the extent to which different companies across the supply chain are involved.
Despite these concerns, Chinese solar was nevertheless heralded by many Western analysts in 2020-21 as a realm of value and opportunity. Margins and market shares were strong, China was a cost-leader, and a moat was emerging in technology R&D. Compared with wind, solar has clear scope to continue getting more efficient – and therefore potentially cheaper – without running into engineering problems where further gains are capped by hard physical laws. As in semiconductors, the intangible value of technological leadership – and the barrier to competitive forces it exerts – is magnified by the rapid pace of change. Comparisons with Taiwan’s burgeoning semiconductor industry in the early 2000s did not appear unreasonable. To gain some exposure to this seemingly exciting theme, Hosking Partners initiated a small position in LONGi Green Energy in July 2021. LONGi has a particular focus on manufacturing solar modules, where it is a technology leader. The company has no manufacturing sites in Xinjiang and claimed to have taken the unusual step of requesting 150 of its suppliers make a written commitment against forced labour. Nevertheless, in September 2021 we began an engagement with the company to encourage greater transparency over the issue.
The engagement spanned background research, 1-on-1 calls with management, and a series of formal letters. We worked to gain a more complete understanding of the regulatory and reputational risks facing LONGi due to forced labour allegations. We asked the company how they managed the risk of forced labour, and how their revenues could be affected by related US sanctions. We encouraged them to offer more transparency through their supply chain, citing examples of good practice drawn from elsewhere in our portfolio. For example, we highlighted how Associated British Foods’ subsidiary Primark has published a supply chain map and associated human rights audit data. Concurrently, we conducted an in-depth assessment of the broader Chinese solar market. The outcome of this work would help our investment team assess whether to expand our exposure to the sector beyond a single position or reduce it.
Despite our efforts, we made relatively little progress determining the scope of LONGi’s exposure to forced labour. It was difficult to ascertain the true nature of the problem in general, let alone attribute involvement at the company-level. LONGi itself made some limited efforts to address our concerns, but ultimately seemed unable to fully engage on the issue as the Chinese government effectively prohibits companies from accepting in the first place the proposition that Uyghur forced labour exists at all. This makes it difficult for companies to be open about the extent of their involvement, restricting them to general statements about being opposed to forced labour in principle. Furthermore, LONGi was unwilling to call into question the activities of their major suppliers – most notably Daqo and Hoshine – many of whom form long-term ‘strategic partnerships’ as part of the industry’s complex JV network. Our inability to encourage greater transparency on the materiality of the forced labour risk – combined with a deteriorating and related supply-side picture described in more detail below – led us to sell the portfolio’s position in LONGi in August 2023.
Capacity at all costs
Our engagement with LONGi on forced labour ran alongside an in-depth examination of the broader Chinese solar market. Despite the rosy outlook posited by sell-side analysts that stimulated our initial interest, since mid-2021 Chinese solar equities have proved to be a disappointment for investors. Our analysis provides some insight into the reasons for this outcome and offers some pointers to the future.
As described above, China’s dominant global market share was built by cornering the upstream production of polysilicon. Aside from a 2022 spike, subsidised Chinese polysilicon prices have remained near or even below cost (7). Meanwhile, upstream capacity continues to be added despite poor returns, leading to a permanent state of oversupply. To keep the upstream producers alive, the government encourages joint ventures with midstream and downstream companies who help fund the capex. IEA data reveals manufacturing capacity running ahead of demand from polysilicon to modules, with the oversupply growing over time. In 2022, capacity was over double demand (see Figure 2). Combined with increasing polysilicon efficiency on a per wafer basis, this has caused polysilicon capacity utilisation to fall from 85% in 2004 to just 60% today (8).
This slack in manufacturing capacity may take years to unwind. Although there are some signs that financing conditions for capacity addition are tightening – producer Tongwei was recently forced to cancel a proposed equity raise to fund further growth – the top eight Chinese producers still have enough cash fully to fund all stated expansion plans through 2026. Meanwhile, marginal demand growth for solar both in China and internationally is showing some signs of levelling off as bottlenecked power grids struggle to adjust to the rapidly rising share of intermittent generation. A recent IEA report suggests around 3,000 GW of global renewable power generation capacity sits idle, unconnected to national grids which have not been sufficiently built-out (9). That is five times the total amount of solar capacity added to the energy mix in 2022. Globally, electricity grids require annual investment to double by 2030 to accommodate the growth of renewables, a significant step-up in the long-run trend (see Figure 3), and a real constraint on the pace of solar demand growth. If we were to adapt the adage ‘to put the cart before the horse’ for 2023, we might say ‘to put the solar panel before the grid’!
Meanwhile, some of the costs associated with solar production may be beginning to re-inflate. The Chinese manufacturing scale that has delivered such remarkable cost deflation has been catalysed by the web of subsidies discussed above, whether in the form of cheap coal energy, forced labour, or low-cost government funding. These are ultimately unsustainable (in several senses of the word), and as such the profits of each industry player need to be discounted heavily in any valuation analysis. Furthermore, if much of the cost reduction achievable via scale is already behind us, then the focus for future deflation switches to materials, cell efficiency, and cost of capital. Promisingly, cell efficiency may eventually double to 50% or more. But such opportunity is tempered by research which suggests a 1% rise in the discount rate implies a 6% rise in the cost of delivered solar power, while materials bottlenecks across the solar value chain continue to tighten (10). Geopolitics also has an impact. US import bans on Xinjiang-manufactured modules have led to the equivalent of an entire year of European demand (50 GW) being held up in warehouses near Rotterdam (11). Return on capital suffers a double whammy from inflated assets and lower future profits as this inventory eventually floods the market. This effect is magnified by the impact of the further 3,000 GW of unconnected spare capacity discussed above.
None of this is to say that solar will not form a critical part of the energy transition. Or that efficiencies will not continue to improve, or that costs will suddenly rise precipitously, or that demand will disappear. However, when we consider the impact of years of Chinese oversupply, such issues are worth considering when we ask ourselves how long such a situation might last, and what impact it might have on the returns of the leading Chinese producers.
We still seem some way from a meaningful cycle of consolidation or capital discipline among the Chinese players. Analysts have been predicting such a cycle since 2012, when the Chinese government suggested it would incentivise industry M&A, but instead the number of large Chinese manufacturers has increased. The Silicon Module Super League – a group of the world’s largest solar module manufacturers – has grown from five founder members in 2015 to seven today, six of whom are Chinese. The story is similar for polysilicon, where Chinese market share is divided relatively equally between six large firms. This oligopolistic set-up could be attractive if capital discipline was maintained and supply tight, but as we have seen that is not the case. In fact, firms are still frequently spending as much on capex than they earn in operating cashflow (see Figure 4) (12).
On the other side of the Pacific, the US was slow to wake up to the fact that an entire industry had been stolen from under its nose. In response, the Inflation Reduction Act (‘IRA’) now promises to deliver the kind of fiscally supercharged investment cycle for US solar that China implemented back in 2011. Furthermore, from August 2024 the US will expand tariffs on Chinese modules to include those assembled elsewhere in Southeast Asia, which to date has been an easy workaround. These policies are undoubtedly politically as well as economically motivated, but nevertheless the prospect of this new avenue of supply has contributed to a derating of Chinese solar equities over the past year. Average P/Es for the top five producers have halved from 31x in 2022 to an expected 16x in 2023 – cheaper, but still not cheap. And while subsidised coal remains a significant cost advantage for Chinese producers, US natural gas is a worthy alternative and is less exposed to the potentially re-inflationary impacts of rising carbon prices (or taxes) in Western markets.
Despite our scepticism regarding the Chinese solar market – for the financial and non-financial reasons we have described – we remain interested in the opportunities posed by increasing solar adoption. However, as we discussed in a recent podcast with Thunder Said Energy’s Rob West (‘Energy Transition: Active Duty?’), it is measurable supply bottlenecks rather than conjectured demand forecasts which most pique our interest. One company in the Hosking Partners portfolio whose investment case expresses this idea is Ferroglobe. One of the largest producers of silicon metal and related alloys worldwide, it primarily supplies sub-solar grade silicon to the chemicals and aluminium industries, and ferro-silicon and manganese to the steel industry. It is vertically integrated from coal and quartz through to the production of silicon metal, and it has a global production footprint of scale which should advantage it in terms of cost, raw material prices and tariffs. It is generally market leader (ex-China) in all its markets. As the West challenges China’s dominance of the solar value chain by means of the IRA, companies like Ferroglobe are well-placed to leverage their existing assets towards the growing market in non-Chinese PV silicon by supplying silicon metal to Western upgraders. Clearly, continued Chinese overcapacity is a risk to Ferroglobe’s pricing power, and we see only early signs that capacity growth may be moderating in China. However, over the longer-term, broader economic pain and over-indebtedness may place pressure on subsidies. This, in turn, could prove more accretive to Western competitors than Chinese incumbents. After all, the former has already been through a cycle of consolidation amidst the last 10 years of Chinese cost-leadership, and Ferroglobe’s healthy balance sheet sets the stage for future returns.
As ever, capex is something to which we pay close attention. On average, it still costs US polysilicon manufacturers over four times more than their Chinese counterparts to add one unit of incremental capacity. Encouragingly, instead of splashing the cash on new assets, Ferroglobe are focusing on deleveraging and delivering returns to shareholders, while exploring strategic partnerships with specialist producers including REC Silicon as the means by which to explore the solar opportunity. Looking forward, incremental silicon metal consumption in North America relating to solar could amount to one third of existing consumption by 2030, implying EBITDA for Ferroglobe of $300m from this sub-segment alone. That compares materially to a 2022 adjusted EBITDA of $795m for the entire business (13). The company also has exposure to developing technology for silicon cathodes in electric batteries, which could prove an exciting source of future profits. Meanwhile, Ferroglobe is free from the regulatory and reputational discount associated with the issue of forced labour in China, which shows no sign of abating in the near to medium term.
Our experience over the past two years investigating – and briefly investing in – the Chinese solar market demonstrates how supply-side analysis, consideration of ESG factors, and active engagement combine in our long-term investment process. As at October 2023, almost 30% of the Hosking Partners portfolio is in sectors closely related to the unfolding energy transition, with a larger percentage indirectly exposed. Currently, much of this is in unloved areas of the market where there has been recent underinvestment and valuations are low. Our confidence in many of these ideas – mining, long-lived conventional energy assets, tanker shipping – is supported by the careful consideration of ESG factors as outlined in this report. Meanwhile, nowhere is the tension between ‘E’ and ‘S’ that is implicit in much of the energy transition more clearly illustrated than in the Chinese solar industry. Over the past decade valuations of renewable energy companies have spiked and are now slowly deflating. This is a bubble we largely avoided, built as it was on unrealistic financing predicated on cheap energy and even cheaper money. Looking forward, we will continue to search for opportunities in emerging energy technologies, where we like companies such as Ferroglobe. These are companies that combine a strong asset-base, regulatory tailwinds and compelling valuation, with a smart capital allocation strategy and improving governance regime.
(1) Thunder Said Energy, ‘Solar Costs: Four Horsemen’, June 21
(2) Thunder Said Energy, ‘Solar Costs: Four Horsemen’, June 21
(3) Maldonado, S. (2020). The Importance of New “Sand-to-Silicon” Processes for the Rapid Future Increase of Photovoltaics ACS Energy Letters, 2020
(4) Murphy, ‘In Broad Daylight’, Sheffield Hallam University, 2021
(5) International Labour Organisation
(6) Murphy, ‘In Broad Daylight’, Sheffield Hallam University, 2021
(7) Thunder Said Energy, ‘Solar Costs: Four Horsemen’, June 21
(8) Thunder Said Energy, Global Polysilicon Production Capacity
(10) Thunder Said Energy, ‘Runaway train: energy, interest rates and inflation’, October 2022
(11) Empire of the Sun, Redburn, Sept 2023
(12) Bloomberg, as of 2nd October 23
7 November 2023
Dark Side of the Sun
A closer look at the Chinese solar industry