Description

Disclaimer: we are currently short JinkoSolar shares and the below represents our opinion.

SECTION I - INTRODUCTION

JinkoSolar (“JKS”, “Jinko”) operates in a fragmented, commoditized, hyper-competitive and deflationary industry—solar module manufacturing. Following a ~250% runup in the stock since June, we believe Jinko represents a highly attractive short opportunity with >70% near-term downside in a base case and more in a bear case scenario. Our view is based on several factors:

• The spike in the shares appears to be tied to three things: (1) the view that the global demand picture is improving and the currently elevated margins can remain so for longer, (2) the widespread belief that a Biden victory in the presidential election represents a significant source of earnings upside for the company, and (3) the company plans to list a subsidiary in China’s A-share market, where solar stocks currently trade at a higher average multiple of earnings, and some investors have speculated that the proceeds of the Shanghai equity issuance will be used to buy back the NYSE-listed ADRs at a large premium. We believe bullish investors are incorrect on all three accounts: 

• (1) We believe that Jinko is over-earning due to short-term mismatch between spot market trends and the company’s reported results. This dynamic is clearly evident when comparing 3Q and 4Q guidance to close competitors, yet the market has extrapolated a very temporary trend. Jinko earnings quality is poor in our view and unlikely to improve. Due to the limited disclosure the company provides, we believe that investors are unduly reliant on the company’s reported adjusted metrics, which are not telling the full story. Industry fundamentals are actively rolling over, and we consider the company’s position in the market mediocre. There is significant low-cost supply entering a market where Jinko’s gross margins still quite elevated relative to recent levels, and a normalization in earnings is already underway.

• (2) If Biden is elected, his plan would likely boost solar demand in the US over the next 5 years, but Biden’s election is actually more of a negative than positive for Jinko. The extension of the current investor tax credit will result in a rapid deployment of modules currently held in inventory by developers, reducing 2021 import demand. As the section 201 and potentially 301 tariffs on solar modules are more likely to be removed under Biden, there is a high probability of over a thousand basis points of estimated margin compression in Jinko’s most lucrative market.

• (3) In discussions with investors, Jinko management has made it clear that proceeds from A-share equity issuance cannot be used to retire ADRs. Valuation gaps between dual-listed H-shares and A-shares can be arbitrarily and sustainably high due to the closed nature of the mainland Chinese exchanges.  There is no mechanism for value convergence between the two securities.

• Our research on the business suggests significant downside earnings variance relative to consensus, driven by pricing and margin compression. We expect 2021 EBITDA of $328mm vs. consensus $458mm, which at 7x EBITDA results in 77% downside from current. 

Mid-case financial overview vs. Bloomberg consensus:

Source: Bloomberg, company filings, our analysis

Mid-case valuation:

Jinko recent share price performance:

Source: Bloomberg

SECTION II – SUPPLY-DEMAND AND NEAR-TERM EARNINGS

1. We believe the market is extrapolating strong earnings growth into the future based on Jinko’s 2Q results, which in our view were characterized by poor earnings quality, and 3Q guidance, which benefitted from a nonrecurring timing effect. We estimate the global solar module market became oversupplied in 2Q20 and will be at least 85GW, or 54%, oversupplied in 2021.

PV Module Manufacturers Meeting BNEF’s Tier 1 Criteria as of 3Q 2020:

Source: BNEF

Below we simply adjust the Bloomberg list for where it is lower than more recent company disclosure and update based on known capacity additions in 2021. We forecast over 200GW of tier one supply today across 40 manufacturers and 142GW today across the top ten, which will be 241GW and 179GW, respectively, in 2021:

PV Module Manufacturers Meeting BNEF’s Tier 1 Criteria, Today vs. 2021:

Source: BNEF, our analysis

This estimate is the lower bound since many other module players are adding capacity as well, and it already represents 85GW of excess supply vs. BNEF midpoint 2021 demand estimate of 156GW. Said another way, the excess supply projected in 2021 from the list above is greater than the entire global demand for solar in any year before 2017. We believe that the industry really entered oversupply in 2Q20, but the choppiness from the pandemic has obscured industry fundamentals, causing margin weakness to be blamed on demand, not supply.

Wafers

Wafers are an intermediate step in the value chain between solar-grade polysilicon (“poly”) and finished solar modules. We believe that the wafer business is the hidden driver of margin for the vertically integrated manufacturers like Jinko. According to Longi’s discussions with investors since 2Q, a large amount of new wafer capacity is coming online before the end of 2020, driving an estimated 500bp Q-o-Q gross margin compression in this product for the low-cost leader.

Below is a mono wafer supply projection, which we believe is conservative. Key rival Longi should easily have over 50% market share in 2020 and 2021 (relative to demand, not total capacity). Longi and Zhonghuan alone will likely have sufficient capacity to meet global demand.

Our Projected Mono Wafer Supply

Source: our analysis

Wafers impact on modules

Below is a graph estimating, based on sporadic company disclosure, Jinko’s clean gross margins vs. the percentage of their total wafer capacity that is monocrystalline (3Q20 midpoint of guidance included). During the period depicted, the industry underwent a rapid shift from the legacy multicrystalline wafer technology to monocrystalline, and Jinko was able to ride that temporary changeover in supply to several hundred basis points of margin expansion. 1Q20 was the peak, following which a wave of new capacity from Longi, Zhonghuan and Jinko entered in 2Q.

Source: company filings and quarterly conference calls

According to BNEF’s illustrative build for a vertically integrated mono module manufacturer based on 2Q spot prices, a manufacturer could earn a 14% margin selling mono wafers externally on the open market, while only a 12.9% margin selling mono modules on the open market. This table is evidence that a normalization in wafer margins should drive a normalization in module margins, something that has not received much attention in the sellside’s analysis of Jinko’s business. Jinko does not sell many wafers externally and thus investors cannot see the internal mechanics of their margin structure.

BNEF Vertically Integrated Mono Module Economics, 2Q20

Source: BNEF, our analysis

If the above analysis is rolled out to 4Q20 to adjust for Longi’s anticipated -500bps of gross margin pressure in wafers and the elevated poly price running through the chain, then module margins even passing through the entirety of the poly inflation will be down almost 150bps. This compares to company guidance of “slightly down”.

BNEF Vertically Integrated Mono Module Economics, 4Q20E Estimate

Source: BNEF, our analysis

In summary, we believe Jinko is facing a steeply oversupplied market for modules and wafers, the bulk of incremental capacity is being added by low cost players (Longi, Canadian, Zhonghuan, and others), and the hidden driver of the company’s dramatic margin expansion is reversing before the end of the year. Investors should not underestimate the magnitude that these margins can give up in a short period of time – the last solar cycle in 2017 saw Jinko’s gross margins fall from 22.1% in 3Q16 to 10.5% just 3 quarters later, and that was during a period in which global demand continuously grew.

Near-term earnings trajectory—what is around the corner?

• 2Q – In the second quarter of 2020, JKS surprised on shipments (+2.7% beat vs. guide), selling prices (+2.1% vs. guide), and gross margins (+90bps vs. guide). Keep in mind they guided on June 15th, so probably had decent visibility into the quarter’s results at the time, but the reported numbers were remarkably resilient. Jinko grew shipments 31% Q-o-Q, and only suffered 11% Q-o-Q ASP decline and 160bps Q-o-Q GM compression, despite a mix-shift towards China in a quarter where macroeconomic uncertainties were at a peak and there were reports of widespread cancellations and order deferrals.

Source: Energytrend, company quarterly earnings release

Source: company filings

• We believe two factors drove better 2Q performance compared to expectations. First, Jinko has an order book which secured some volumes at prices above the collapsing market price—there is evidence of their order book based on this snippet from the 2019 20-F: “Pursuant to our order book of 2020, we are well positioned with 55% of expected solar module shipments for full year 2020 secured as of March 31, 2020, compared to 65% of total solar module shipments for full year 2019 secured as of March 31, 2019.” Management has shared that purchase orders frequently include a down payment, and typically the contract has a force majeure clause. In the second quarter, Jinko experienced an impact in residential, but the utilities were not cancelling or delaying orders. While no doubt some resi customers (only 10% of the total US volumes) were able to move out shipments or get them cancelled completely, it seems safe to assume that most contracts were fulfilled as written given the high order book coverage as of 3/31. Note that with 25% of the year already gone, Jinko’s 55% coverage rate implies a weighted average contract duration of just over 3 months. This 3-month lag would have the effect of dampening the 2Q ASP decline, although orders written subsequent to March will weigh on achieved prices in the second half of 2020.

• The second factor is the mix-shift to China in the quarter. Roughly 8-10GW of projects missed the 2019 commissioning deadline (12/31), and had until 6/30/20 to commission and receive a slightly diminished feed-in tariff. We believe that Jinko quickly diverted volumes into China, which would have been dilutive to GM but allowed the company to outperform their shipment guidance. The negative GM impact may have been somewhat offset by the contract mix effect described above. We believe that the last-minute sales rush in China is the reason for the large net working capital headwind in 2Q, where Jinko burned $100mm of CFFO despite reporting $119mm of EBITDA. Accounts receivables alone caused ~$70mm of cash flow headwind, which could be due to Jinko rushing volumes out the door right before the quarter ended and before the cash could be collected. Chinese customers are known to have considerably longer accounts receivable turns because they buy on credit and depend on the frequently delayed government subsidy program for cash to repay the equipment manufacturers, whereas western customers pay upfront deposits which negates the working capital use of cash.

• 3Q – moving to 3Q, the higher-priced contracts would be expected to roll off, causing Jinko to experience significant sequential pricing and gross margin pressure, with the potential for some cost pressure as the upstream polysilicon disruptions work through the P&L. 

• Instead, while the midpoint pricing guidance was considerably weaker than expected at -9% Q-o-Q (confirming our suspicions about the contract lag), the margin guidance was flat Q-o-Q as unit costs were guided down 9% as well. This is difficult to fathom in a quarter where spot poly prices rose ~40% Q-o-Q, and closed the quarter a whopping 83% above the average level in June, not to mention other important inputs like silver paste were up ~45%. 

• The company’s explanation is that they will not see the impact of higher input prices until September and the brunt of the impact will fall in 4Q. Management also shared that they did not sign many contracts in China in 2Q (when the market price was at trough) so they are benefitting from higher average pricing in 3Q. In other words, Jinko is expensing the last of its lower-cost polysilicon inventory in 3Q, depressing its manufacturing costs, while benefitting from the sequentially higher prevailing market prices in 3Q (up to +$0.02/W). This mismatch (prices that are passing through a cost not yet incurred) is purely a GAAP phenomenon, since Jinko’s cash outlay on poly will have stepped up already. This is also confirmed by management, who said that the net working capital headwind would persist (our expectation is that the headwind will be from inventories in 3Q vs. A/R/advances in 2Q). 

• 

Source: company quarterly earnings calls

We believe investors less focused on the details of the earnings trajectory have extrapolated the headline guidance forward into 2021, whereas a deeper look reveals earnings quality concerns and timing-related effects. The underlying picture is quite challenging for Jinko and the broader industry, as early as 4Q20:

• 4Q – We believe 4Q gross margins will be disappointing. There are a number of factors in our view combining against Jinko: 

• (1) new wafer and module capacity will have been added, estimated above to be a 140bp headwind based on our analysis above

• (2) the higher poly prices will finally show up on the P&L, which could be >100bps of headwind based on CSIQ commentary. This is straight from the CSIQ 2Q transcript: “So altogether, I think when it comes to modules, the impact is like $0.02/W. And so, however, in Q3, for pricing, for module price up, we try to reach price for module, but does not mean we can reach 100%. So I would say, we still lost at least 5% margin. That's probably the rough number.”

• (3) the USD has weakened against the RMB to become a significant headwind, which we believe implies 80bps of headwind Q-o-Q at spot assuming 50% hedging:

Estimated Impact of FX on 4Q20 Gross Margin

Source: our analysis

• (4) the mix of China installs (which is lower-margin) will be much higher, which is the typical seasonality and has driven an average 4Q GM decline of -270bps since 2015, and 

• (5) there is simply no credible reason why Jinko should not experience the same margin headwinds that peers are experiencing already, even if there is a temporary mismatch in 3Q. 

• On the 2Q call, an analyst asked whether a 15% GM was likely in 4Q, which would be -300bps vs. 3Q midpoint of guidance. Management said that was too conservative and the margin would be “slightly down.” Our work laid out above would suggest something on the order of 300bps is more than likely even if operating leverage from greater production volumes provides a partial offset. One or two of the points above could be wrong and there is still sufficient headwind to drive a material stepdown in 4Q margins, which we believe should persist into 2021.

• To sum up, here is our best estimate for the key moving parts quarter by quarter:

Source: our analysis

SECTION III – BLUE WAVE IMPACT ON JKS BUSINESS

2. Investors mistakenly believe that a democratic sweep of the Senate and Presidency is a material positive for JinkoSolar, when in reality there are likely significant headwinds to near-term volumes and margins if the blue wave outcome occurs.

Biden – renewables policy and implications for Jinko?

• It would unrealistic to assume that some amount of the recent JKS enthusiasm is not coming from US investors handicapping ever greater odds of a Biden victory and a large democratic Senate majority next month. US-listed solar supply chain stocks, which usually trade with a high beta to the market, have recently outperformed even on days when the market is down big. So what are the likely policy changes that would benefit Jinko? The main one being openly speculated about is an extension of the ITC for solar at 30% for another five years (or longer), with an added inclusion for battery storage. There is also likely a clean energy standard, which would require utilities to reduce carbon generation annually. BNEF anticipates that even in a blue wave outcome, they do not expect a major policy change to impact 2021 demand, but it would be a significant (though hard to quantify) boon for development in 2022 and beyond. In any case, the key form of benefit for Jinko and the other manufacturers would be incremental volumetric demand.

• There is logical speculation that Biden would eliminate the section 201 tariffs Trump implemented in 2018, which were ostensibly put in place to stimulate local production of solar cells and modules, but caused US solar prices to approach double the levels seen anywhere else in the world and mainly benefitted FSLR, the only domestic manufacturer of note. The tariffs are structured to step down 5% a year over four years. They began at 30% in February 2018 and so they are currently 20%, stepping down to 15% in 2021 and 0% in 2022. If Biden so chose, he could decrease module costs 15% in a day with the stroke of a pen. There is an added wrinkle in the 201 tariff allowing bifacial panels to be imported with no tariff at all, so the US market has shifted towards this technology for projects where it is practical (ground-mounted utility-scale projects, mostly), but the remainder of the market continues to import monofacial modules at a 20% rate. It is worth bearing in mind that Jinko has 400MW of capacity in Jacksonville, FL, so they would be harmed by the removal of these trade barriers. BNEF estimates there is a huge surplus of capacity that can serve the US demand at an all-in cost below $0.30/W (current spot price is mid to high $0.30s/W). JKS SEA capacity would be in the volume listed between FSLR and the Hanwha/Jinko US capacity below:

PV Module Supply Curve for U.S. Market, 2020

Source: BNEF

PV Installation in the U.S. by Applications and by Module Type, 2020

Source: BNEF

• The other tariffs in force today are the section 301 tariffs against Chinese imports, a flat 25% on a range of goods originating in China. Jinko has gotten around this by shipping wafers to their cell and module manufacturing facilities in Malaysia, which had 3GW of capacity at the beginning of the year. So the current capacity capable of serving the US market is 3.4GW, 400MW of which is completely tariff-free and 3GW of which is paying only the 20% section 201 tariff on monofacial shipments. We believe the removal of the 301 tariffs, in the case of a Biden rapprochement with China, would be a complete disaster for Jinko since the price would rapidly normalize to the global average and the most lucrative market in the world would cease to be so.

• One final point on the implications of a Biden victory. The US market has been massively prebuying modules in conjunction with the stepdown in the solar investor tax credit (“ITC”). BNEF tracked 21GW of imports into the US in 2019, compared to only 11GW of installations, meaning some large number (~9GW) is sitting in inventory and will reduce future procurement of modules in the 2021 timeframe and beyond:

Imports and Installations of Solar Modules to the U.S., 2019-2020

Source: BNEF

• The extension of the ITC for five+ more years would have the side-effect of rendering that 9GW almost worthless, since these are now obsolescent modules which would need to be installed immediately, reducing Jinko’s apparent demand in 2021 by almost half.

• So the Biden win is likely good for solar shipments over the next five years, but we believe it is a significant negative for module shipments in 2021 and a significant negative for module margins in the scenario where tariffs are removed. It actually seems to be a worse outcome for Jinko’s earnings than a Trump win, since Trump would be more likely to sustain the trade barriers that have hitherto allowed Jinko over-earn in the US market.

SECTION IV – CHINA SUBSIDIARY LISTING PLANS

3. Jinko’s ADRs are unlikely to converge up to today’s prevailing mainland China valuations, whether or not the company is successful in its announced subsidiary listing plans. The IPO proceeds will not be used for buying back ADRs.

There are two variants of the bull argument we have heard regarding the above corporate initiative to access China’s capital markets. The first is purely around cost of capital, while the second is more about closing the valuation gap between the US and China. 

1. Jinko’s plans to list in China dramatically reduce the company’s cost of capital such that the through-the-cycle ROIC (~6%) is now sufficient to create value through the cycle

This is part of the stated rationale for the listing and if investors think back, was also the reason Jinko listed on the NYSE in the first place way back in 2010, as they sought to ride the hype of a prior solar cycle into a premium equity valuation courtesy of US capital markets—which then derated the stock to the discounted levels at which Jinko has traded ever since. Leaving aside for the moment the likelihood of such a derating occurring in the A-share market (probably the most volatile equity market in the world) at some point in the future, or the execution risk inherent in effecting a quasi-dual-listing NYSE/SHSE structure that has never been attempted, the bare premise of the listing makes sense. The size of a potential A-share IPO is unknown today, but assuming something like a $1bn / RMB 6.8bn offering, it does not seem to produce a WACC that would create a lot of value over time (it is still only approximately equal to through-the-cycle ROIC), although perhaps they could continuously issue equity in China until a better result was achieved, if the market would bear it.

Hypothetical Pro Forma JKS Cost of Capital

Source: our analysis

So even in hypothetical scenario where the Shanghai listing goes off without a hitch, the recent dramatic jump in the ADRs does not seem merited if this is purely about cost of capital, since the company is going from an abysmal profile to a merely neutral one.

2. The relisting will result in Jinko’s ADRs either rising to the level of prevailing clean tech valuations in the A-share market or being taken out at a significant premium

A rumor was circulated in August that Jinko was seeking investors to take the company private before re-listing on one of China’s mainland exchanges – a strategy similar to what Trina and JA Solar have done. Instead of taking this route, however, the company has followed Canadian Solar’s subsidiary listing strategy, the first of its kind to be attempted, as far as we know. More clarity from both Jinko and Canadian on the background of these processes, the level of prior engagement with local regulators, and the different regulatory hurdles to be cleared would certainly be appreciated.. More Importantly, Jinko management has shared with investors that they cannot retire the ADRs with the proceeds from the Shanghai listing, so the rumored take-private/buyout appears to be out of the question.

As far as a market-driven convergence between ADR and A-share valuations, the markets for the dual-listed equities appears to be completely disconnected.  The domestic holders are largely unable to buy the international listing, and international investors are unable to short the A-shares, so the typical free market price arbitrage mechanism is not able to fully function.. This is from a UBS note regarding valuation gaps between the A-share and international markets:

• “Fungibility – dual listed shares being fungible tend to bring both lines to near parity…[it should be pointed out that JKS ADRs are not fungible with the Shanghai-listed security]

• Accessibility – for developed markets, the ability for investors to have access in trading both lines of shares will ultimately bring both lines close to parity.  In China, international investors (who I view as more fundamental in nature) can only access A-shrs via QFII and StockConnect, so it is relatively restrictive 

• Investor mix - foreign investors can only access Chinese A-shares via StockConnect or QFII.  They currently own ~7.3% of freefloat of A-shrs at the moment.  That leaves the rest being owned by local pensions, asset managers and retail.  The multiples local investors pay may be very different than how international investors may pay for the same company.  Generally, A-shr investors are happy to pay higher multiples for stocks with relevant thematics and growth.”

Below is a chart showing the volatile, seemingly random relationship between the Hong Kong and mainland China share prices of dual-listed stocks (red-brown line below). The disconnect has persisted for years and is currently testing all-time high spreads.

A-H Premium for Dual-Listed Equities

Source: UBS

In our experience with another dual-listed stock, investors were actually shorting the international listing initially and buying the A-share, despite the A-share trading at nearly double the level of the international shares, perhaps as a way of hedging out everything but the technicals and momentum of the Shanghai STAR exchange, or as a natural consequence of international short-biased investors having only the H-share borrow available. While it is unclear whether this will happen with Jinko, at the very least investors should not expect a convergence in premium/discount between the two securities to occur organically.

SECTION V – FORWARD EARNINGS AND RISK-REWARD

4. To summarize the discussion above – 2Q reported and 3Q expected earnings trends convey in our view a misleading picture of the resiliency of a company that is, at the end of the day, a commodity manufacturer in an oversupplied market. We believe positive investor sentiment around a change in US renewables policy is overblown to say the least, and the logic of a massive upward convergence in the value of the ADRs to mainland China trading ranges is flawed. We conclude below with a view on 2021 earnings and our target valuation for JinkoSolar’s ADR.

2021

• The company raised year-end 2020 projected module capacity to 30GW and said something in that range is a fair assumption for 2021 shipments, which is +58% vs. 2020 midpoint. This is a big number, but volume growth is ultimately dictated by demand, and a bigger driver of the economics in 2021 is the magnitude of pricing declines. Over the last four quarters, Jinko’s reported Y-o-Y ASP decline has averaged just -3%, compared to the Energytrend average ASP decline for mono modules over the same period of -22%.

• Some of the outperformance has been regional mix (we estimate +630bps in 2019 and +130bps in 2020E), due to Jinko’s rapid growth in the premium-priced US market. But as discussed above, we believe this will reverse in 2021, and -300bps seems like a conservative base case (assumes they add 1GW of US supply in 2021 into a market that is already heavily oversupplied). This regional mix assumption does not include tariff elimination, so there is some more downside relative to spot if that is the ultimate result, and it does not include Japan falling apart (BNEF projects double-digit average declines over the next two years, in a market that is possibly even more lucrative than the US).

• Some of the recent pricing outperformance appears to be product mix, as Jinko has been in the top handful of players spearheading a drive towards larger-size wafers today, and in 2019/early 2020 was benefitting from the rapid shift in industry demand to monocrystalline wafers, where the company was concentrating its investments. Both of these are ultimately a commodity as much as anything else, and investors should not expect this to insulate them from pricing pressure in 2021. In fact, since all of the capacity being added is targeting the newer, faster-growing products, these are likelier to experience more near-term overbuilding and deflation.

• Some is contracts, which, as discussed above, approximate market pricing on a short delay. That could result in another ~300bps of pricing normalization looking at 2021 vs. LTM (after all, the market was down 30% in 2Q vs. JKS down 10%, suggesting 500bps annualized over-earning, but it was not all from contracts).

• All-in-all, we believe Jinko will catch up to the market, underperforming the spot price index by ~6ppts (3ppts from regional mix, 3ppts from contracts resetting lower or product mix deterioration). In a typical year the market price declines 17%, but there are unprecedented levels of supply entering the market for both wafers and modules so it could be worse. If the market price falls 17% and Jinko is 6ppts worse, then the company will experience 23% pricing declines, with further downside potential from the US pricing normalizing to global averages, FX headwind, etc. This would yield a pretty negative result, since unit cost reductions over time for these businesses should not be better than mid-teens. As a result, we project significant margin compression in our earnings model.

Source: company filings, our analysis

• With mid-case ’21 EBITDA of $328mm, and a 7x EBITDA multiple, which seems more than fair looking at both the recent trading ranges (before this spike) and the generally low quality of the EBITDA number they report as a proxy for cash flow, the resultant price target is $13/sh, -77% from current.

SECTION VI - APPENDIX

Solar modules are the main component of solar hard costs, thus module cost declines are the main driver of solar levelized cost of electricity (“LCOE”) savings. As such, it is unsurprising that module pricing per watt has consistently declined over any significant historical timeframe. The pricing CAGR from 2012 to 3Q20 was -17%:

Source: Energytrend

• The R-square between the chart above and Jinko’s reported quarterly selling prices since 2015 is 0.970, which is evidence both of the lack of differentiation in the module category and the validity of using market benchmarks as a proxy for Jinko’s own module pricing.

This deflationary trend has been driven mostly by the manufacturers adding scale—spreading fixed costs over an ever-wider volume base—and subtle tweaks to technology leading to larger and more efficient panels. These drivers of unit cost reduction both require a high level of continuous capital investment to keep up with competitors who are similarly adding capacity. As a result, through-the-cycle returns for solar companies are unimpressive. Below is JinkoSolar ROIC since the financial crisis:

Source: company filings

This is an emerging markets manufacturing business, and while earnings per share is somewhat meaningless due to the lack of any observable relationship with cash flow generation, the average forward P/E since Jinko even had positive expected earnings has been 4.75x, implying a cost of equity (and by extension a WACC) well in excess of the company’s ~6% ROIC. The chart above actually overstates industry returns due to “survivorship bias” since Jinko is one of the relatively few solar companies to survive this long without undergoing a debt restructuring. The “average” solar manufacturing ROIC would capture the many firms that have been forced to exit the industry or were delisted due to poor returns and would therefore be significantly lower than Jinko’s ~6%. Examples from the recent past include:

• JA Solar (“JASO”, which traded down from >$100/ADR to $7.55 before being restructured in a management take-private in July of 2018)

• Trina Solar (“TSL”, which traded down from >$30/ADR before also being taken private in March of 2017 at $11.60)

• Yingli Green Energy (“YGE”, which is a penny stock today after peaking at over $350/ADR)

The continuous and rapid expansion of wattage per module, driven in turn by persistent reinvestment of any operating cash flows (or more, likely, proceeds from capital issuance), has the other consequence of causing rapid technological obsolescence, such that a module produced today must be quickly sold or its value in several months is significantly impaired. Working capital is a persistent headwind as a solar manufacturer adds scale to reduce unit costs, and manufacturers often drive sales by extending credit to poorly-capitalized developer counterparties. Below is Jinko’s net income vs. free cash flow (the only year of positive FCF since the company has reported financials came during a year of negative net income):

Source: company filings

In Jinko’s case, the reported EBITDA number is also fairly useless in understanding the cash flows of the business – in years in which Jinko even had positive EBITDA, the average conversion of EBITDA to operating cash flows is -3.7%:

Source: company filings

These long-term features of the solar module manufacturing industry–double-digit price deflation, ROIC < cost of capital, and persistent cash burn—are structural and unlikely to change anytime soon in our view. It is a tough business where operators must run to stand still and thus should garner a low valuation multiple. Historically, we believe that the best time to short module manufacturing stocks is when gross margins are at peak since that is when module manufacturers are most incentivized to overbuild new capacity.  The lack of industry consolidation suggests this trend is highly susceptible to repeat. 

Demand – China

• China – China accounted for 33GW of demand in 2019 (several GWs of projects auctioned in 2019 were pushed into 1H20), and BNEF has midpoint forecast of 40GW in 2020 and 51GW in 2021. There is significant near-term risk to the 2020 volume forecast, since developers and manufacturers are playing a game of chicken over whether the 3Q polysilicon inflation will be passed through or whether developers will simply wait for lower prices to come along. The standoff is with respect to $0.015-0.020/W of increase, which is +8-11%. Our research suggests 8-9GW of demand is at risk based on the trend in module prices through year-end, some of which would be embedded in the BNEF number. They, JKS and CSIQ have said that they hope to recover most but not all of the inflation. Whatever of the 40GW does not fall in 2020 will simply be pushed into 1H21, so it is not a significant concern—the magnitude of feed-in tariff foregone from a later commissioning is not enough to incentivize developers to pay the higher module prices prevailing today.

• Central government’s new five-year plan

• So far it seems that an announcement from the central government regarding the new five year plan for renewables development is likely in 4Q20, and the draft proposal looks like an average of ~60GW per year of deployments, which probably ramps from ~50GW in 2021 to >60GW in 2025. So far the central planned solar scheme has been a failure in many ways, characterized by late subsidy payments, grid curtailments due to congestion, lengthy interconnection queues and resultant financial distress among private developers as project IRRs have deteriorated.

• Feed-in tariff going to grid parity?

• The move to grid parity has been met with enthusiasm by industry participants, both as a sign that solar LCOE is now below the marginal cost of coal generation, and because of the issues that plagued the feed-in tariff scheme. We believe the change is not as transformational as it may sound. Rather than resembling a western merchant power market, there are still implicit subsidies granted by the central government to encourage solar development, including grid interconnection priority over coal, a 30-year PPA of sorts that benchmarks to the coal RMB/kWh price, the tax, credit and land concessions that were in place before, and of course the fact that most utility-scale solar development is done by SOEs that are ultimately answerable to the central government. The annual auctions, the source of much of the historical lumpiness in the Chinese market, will occur as before, except now the developers are competing for grid slots instead of subsidies.

• Bloomberg takes the view that demand is high and the economics are in the money for grid parity, so there should not be a big dip in demand next year. But no one really has any visibility since these projects will not even be allocated before 2Q21 and will not commission until 4Q21, preserving the typical mad-rush seasonality that characterized the prior regime. As a counter-argument to Bloomberg’s view, the last time they froze subsidies was 2018, and that did not end well (installations fell 16% Y-o-Y).

• 2021 and beyond

• The BNEF projections represent a good base case, but we believe there is more downside than upside risk given the change in incentives, the current lack of visibility into the project pipeline, and the fact that key sources of historical volume disappointment—late year deployment rush, limited interconnection capacity—have remained. 2021 demand of 51GW gets them back to near peak historical development, and then BNEF assumes a modest slump to 46GW in 2022 in the mid-case. So nothing too exciting here.

• The U.S.

Is the US the hidden profit center of Jinko? How much does it matter?

• The company’s disclosure does not distinguish the profitability of the US market, and it is an increasingly important question since Jinko could (a) benefit immensely from a volume bump in the US if it represents a disproportionate amount of earnings, or (b) suffer a significant blow if the tariffs propping up earnings are removed.

• Here is what is known based on company commentary – North America represented 8% of 2018 shipments, 17% of 2019 shipments and something guided larger, ~20%, of 2020 shipments. In 2018, sales to North America were 11% of the total and in 2019 were 25% of the total. It is not easy to say exactly where pricing will shake out in 2020, but in our mid-case, JKS will derive around 30% of revenues from the US in 2020. The company appears to have quickly retooled following the imposition of the 201 tariffs in 2018 and has grown US revenues by ~$500mm in each of the two years since, likely because Chinese peers were not able to get SE Asian (i.e. 301 tariff-free) capacity online as quickly as Jinko did. The charts below demonstrate that SE Asian imports completely dominate the mix of US imports, led by Malaysia:

U.S. Imports of PV Modules

Source: BNEF

• Based on management commentary during 2019, Jinko was earning margins significantly above corporate average in the US, we estimate up to 30%. Those margins have likely taken a step down in the second half of 2020, but we assume the full year averages out around 27%.

• As the BNEF analysis makes clear, 2021 may not be a big year for US module imports, since almost half of demand will be met by inventories already sitting on-hand. But even assuming Jinko manages to ship another gigawatt of modules Y-o-Y, or 5GW, into the US in 2021 (and bear in mind they only had 3.4GW of capacity that can economically serve the US, last anyone knew), regional mix will be a significant headwind in 2021 (around 300bps impact on selling price change Y-o-Y), and they will have trouble growing the gross profit dollars of the US business in a scenario where prices and margins are normalizing to the levels of the rest of the world:

JinkoSolar Projected Regional Shipments, Revenues and Margins

Source: company filings, our analysis

Catalyst

Earnings miss

Tariff removal