Toshiba Corp 6502.T S W
April 20, 2016 - 11:44am EST by
2016 2017
Price: 234.00 EPS -92 11
Shares Out. (in M): 4,238 P/E -2.6 21.7
Market Cap (in $M): 9,500 P/FCF 0 0
Net Debt (in $M): 15,500 EBIT -4,000 900
TEV (in $M): 25,000 TEV/EBIT 0 28
Borrow Cost: General Collateral

Sign up for free guest access to view investment idea with a 45 days delay.

  • Technology
  • Japan
  • Conglomerate
  • Accounting restatement


Thesis Summary:

Toshiba, the Japanese industrial conglomerate, is my highest-conviction short idea over the next 12-24 months. While a year-long accounting scandal and its effects have led to significant underperformance over the last year, the recent short covering-led rally has led to an excellent entry-point for what I consider still the early innings of a multi-year, value-destroying journey. Despite recent asset sales at better-than-expected terms, the balance sheet remains woefully over-levered, stuffed with intangibles, and at high risk for short and medium-term impairment. At the same time, the forward outlook on the post-restructuring business portfolio is at worst case dire, and at best case far less bullish than management guidance suggests. Under a variety of valuation methodologies, Toshiba equity looks to over-valued by 40-70%, with downside tail risk of full impairment in a couple of years.  Base-case valuation in line with key comps like Hitachi – after needed recapitalization via equity issuance - suggests a stock price of ~80 JPY/share, or 65% downside from here, while even if Toshiba hits ambitious two year targets, a similar approach suggests the stock shouldn’t trade north of 150 JPY (~35% downside). Further, this idea is highly actionable as the stock is extremely liquid, trading ~$120-150mm ADV, and is an easy and cheap borrow.

Key thesis points:

1.       Asset sales have not solved the leverage problem and balance sheet quality is as worse as it has even been

2.       Go-forward core businesses face large challenges

3.       Management mid-term guidance assumes beyond-heroic profit assumptions across many businesses

4.       Valuation on most all metrics is wildly divorced from direct comps and fully discounts a massive earnings recovery

Let’s take each point one by one.

Asset sales have not solved the leverage problem: Without delving too deeply into the company’s past, Toshiba – for many years before the accounting scandal came to light – had operated with too much leverage and a balance sheet stuffed with intangibles. This was a function of debt-funded acquisitions (Westinghouse, in 2006, for $5.4bn; Landis & Gyr, in 2011, for $2.3bn; IBM’s POS business, for $0.9bn, in 2012) at peak-cycle prices - cumulatively, these added 600bn JPY in goodwill and at least 200bn+ in other intangibles to the consolidated balance sheet. Additionally Toshiba had large (300bn+ JPY) DTAs, a function of chronic losses in legacy consumer electronics businesses (TVs, PCs) as well as commoditized semiconductor segments. The cumulative pressure of a number of bad acquisitions, coupled with a bit of bad luck (the Fukushima nuclear disaster in 2011), were key factors in promoting the culture of accounting fraud that unraveled last year.

The key point is that pre-accounting scandal, Toshiba’s equity ratio – a broad measure of balance sheet leverage, net assets (book equity) divided by total assets, that Japanese companies tend to focus on – was quite low, hovering in the 15-20% range from FY 2011-2014. Due to the preponderance of intangibles, tangible common equity has actually been NEGATIVE for many years, with Intangibles/Equity ratio hovering in the 120-140% range during the same time frame.

Of course, the accounting scandal demonstrated that past profitability had been massively over-stated across all divisions (by 220bn + over 5yrs), and necessitated the impairment of a whole host of DTAs (~300bn), along with restructuring costs/impairments to fixed assets in Energy, Nuclear, Semiconductor, and Consumer Electronics businesses totaling ~220bn in FY15 alone. Interestingly, only ~70bn of goodwill – related to the IBM PoS acquisition – was impaired, and nuclear/Landis & Gyr intangibles have not been touched (yet).

By 3Q15 (Dec’15), with a good chunk, but not all, of the aforementioned impairments taken, Toshiba’s balance sheet had deteriorated to an 8.8% equity ratio, 235% intangibles/equity ratio, and adjusted net debt/equity ratio (including unfunded pension) of 232% – highly stressed levels, and this all DESPITE Toshiba having already raised ~200bn in cash from listed equity stake sales in the preceding two quarters. With the public equity markets closed to secondary issuance (Toshiba is on a TSE ‘watchlist’ due to the accounting fraud that won’t be lifted until September at earliest), the company decided to sell its ‘crown jewel’, the Medical business, and garnered a king’s ransom of ~690bn in gross proceeds (and ~520bn in net profits and approximate net cash), via selling the business to Canon.

I have a number of problems with the price and transaction (Canon basically paid 25x EV/EBITDA for a business showing negative top and bottom line growth the last three years, when other companies in the same industry trade at low teens EBITDA multiples), but the salient point is that despite this windfall gain, and also the subsequent sale of their home appliance business to a Chinese competitor for ~54bn, Toshiba’s pro-forma balance sheet remains in a precarious position and still desperately in need of recapitalization:

-          As of Dec’15, Tosh had 528bn of equity, 1224bn of net debt, 1913bn of adjusted net debt (including pensions and guarantees for third-party debt), equating to 8.8% equity ratio, 232% net debt/equity, 363% adjusted net debt/equity, and 235% intangibles/equity ratio

-          Pro-forma Mar’16 for all announced transactions and estimated 4Q losses, Tosh will have ~618bn of equity, 854bn of net debt, 1673 of adjusted net debt, equating to a 10.4% equity ratio, 140% net debt/equity, 271% adjusted net debt/equity, and 201% intangibles/equity ratio

When looking at the equity ratio, a quick rule of thumb (in Japan, at least), suggests that below 20% is too levered; below 10% is far, far, far too levered. As a general principle Japanese banks are extremely uncomfortable lending long-term funds to corporations when the equity ratio falls below 10% - especially when that company – like Toshiba – has large extant debt and long and tenuous working capital cycles. Recall also that both times Sharp, the pre-eminent distressed Japanese corporate, fell below 10% equity ratio in the last four years, either a large equity issue or a debt for equity swap followed in the subsequent quarters.

The main reason for the lack of absolute improvement is a huge increase in unfunded pension liability (~130bn QoQ increase as guided by the company). Toshiba has one of the largest unfunded pensions (559bn as of 3Q) in Japan, and as of Mar’15 assumed a discount rate of 1.5% on future obligations. With Japanese government bond rates now 10bps NEGATIVE on the 10yr and just 300bps for the 30yr – along with the correction in the equity market hurting the asset side of the book – it is highly likely that the additional 130bn in charges taken in 4Q is just the start. Goldman Sachs, for example, estimates that every 0.5% change in the discount rate increases the unfunded liability by 180bn – that alone is ~30% of pro-forma book equity at year end.

Why the balance sheet pain could get even worse: Of course the elephant in the room during this discussion has been the massive remaining goodwill balance (~680bn as of Dec’15). As mentioned, the only goodwill hit taken thus far during the restructuring process was a 70bn writeoff related to the IBM PoS business. This is surprising for a few reasons:

-          The acquired Westinghouse business has underperformed targets for years, and indeed has not won a single new reactor order since Fukushima (2011);

-          Westinghouse actually impaired its goodwill at the subsidiary level TWICE (in 2012, and 2013) by a cumulative $1.5bn (~180bn JPY); but Toshiba did NOT impair the consolidated balance sheet on either occasion (nor did it disclose the impairments to the market until prompted by a whistleblower leak late last year), suggesting instead that the ‘overall’ valuation for Westinghouse had risen despite specific sub-segments (new construction) underperforming (??);

-          Toshiba actively impaired a number of ongoing nuclear projects both this year and last year, which likely included Westinghouse business (though Tosh did not disclose the impairments on a project by project basis)

In any case, recent reports in the Asahi newspaper suggest Toshiba is considering up to 200bn in goodwill impairments related to the Westinghouse business as part of its year-end (Mar’16) accounting. This is likely an opportune time to finally come clean, given:

 a) the sale of Toshiba Medical has reduced, for now, the critical solvency questions surrounding the parent accounts;

b) as part of the accounting scandal fallout, Toshiba will move to new auditors from Apr’16 – so it is likely the new accounting staff is pushing to fully clean house before taking over the books (the former auditors, E&Y ShinNihon, were massively fined for being negligent for many years and the treatment of Westinghouse intangibles was a major topic therein); and

c) US government departments, apparently the DoJ and SEC, have subpoenaed documents relating to Toshiba’s inappropriate accounting, perhaps related to the fact that Westinghouse is a US-based (Pennsylvania) corporation and Toshiba maintains a US ADR listing.

As such you would think the parent Toshiba has multiple, good reasons to further impair the business now.

The earlier discussion on pro-forma leverage does NOT include any further intangible impairments. A 200bn impairment as is being discussed (still only ~30% of total goodwill, and just 60% of Westinghouse-specific goodwill) would constitute a ~33% hit to pro-forma equity (from already stressed levels) and take the equity ratio back down to all-time low distressed levels, around 7.5%. And this doesn’t even account for the possibility for further writedowns in other parts of the goodwill (Landis & Gyr, to be discussed later), or further fixed asset impairments in remaining, problematic business units (HDDs, PCs); or further mark-to-market losses in the unfunded pension as rates in Japan compress ever lower.

In any case – even in the best case outcome of no further writedowns at all, Toshiba’s balance sheet remains so levered and of such poor quality that a large equity offering is basically a matter of time. The company has stated they wish to return to 1000bn of book equity (vs ~620bn pro-forma as of Mar’16) and 100% debt/equity ratio (they conveniently ignore unfunded pension from this calculation). But getting to 1000bn book equity, having sold the crown jewel business and with little left to sell at a premium to book value (and with more book-impairing losses to come), seems an impossibility without a large equity offering. Indeed, the company may have presaged this during a recent business plan (Mar 18th), when Toshiba labeled the FY16 year (ending Mar’17’) as a “Return to the Capital Markets” year. Frankly, I anticipate a large (300bn-400bn) equity offering as soon as practicable, probably later in the year once the TSE warning designation is removed from the stock and the company can claim to have returned to ‘normalized’ accounts.

Some may argue that the company has ample liquidity and could delay an offering indefinitely until business performance recovers. While this is certainly a risk, that approach is complicated by the business imperative to return to an IG rating (Toshiba was recently downgraded to junk). As a large chunk of Tosh’s go-forward growth strategy is aligned with the long-term, capital-intensive, nuclear construction business, it will be extremely difficult (nay, near impossible) to win meaningful new business on attractive terms when it cannot provide the financing during the bidding process for new projects (we will touch more on this point later). Suffice to say, there are compelling business rationales, and not just the balance sheet rationale, to try to delever quickly, even if this means significant dilution.

Timing wise, this would probably occur after 1H’16 accounts are reported, thus, October/November in the early case, or maybe early 2017 in the late case. Our subsequent discussion of valuation will take into account the near-certainty that share count will increase, and value the stock on a pro-forma, assumed-dilution basis given this reality (something I don’t believe the street realizes fully).


Go-forward ‘core’ businesses face massive challenges: in addition to the horrid balance sheet and prospect of significant dilution, I love this short because the go-forward business portfolio that Toshiba has assembled looks to face significant challenges. Despite the optimistic outlook for recovery painted in recent management meetings, the company has historically been a very poor forecaster of its own business (even when it was cooking the books) and some of its specific unit forecasts are fairly incredulous. After a bit of background, let’s take a look at three specific problem areas: memory, storage, nuclear, and smart meters (Landis & Gyr). Collectively these ‘problem’ segments will comprise ~45% of Toshiba’s annual revenues going forward.

Quick background: one of the many outcomes of the accounting scandal has been a reshuffling of Toshiba’s business portfolio. On the one hand, Tosh is abandoning many long-held ‘sacred cow’ businesses like TVs, PCs, home appliances; home appliances has been sold, TVs have been minimized/reduced to runoff, and PCs is apparently going to be merged and moved off-balance sheet (though this remains a source of risk as the three-way merger with Vaio and Fujitsu appears to have fallen apart). At the same time, some of the more commoditized semi-conductor businesses (LSI and Discretes) have been radically restructured and will not be needle-moving going forward. Instead, Tosh is going to focus on three key pillars going forward: Storage/Memory (~25% of group sales, mostly NAND memory as well as HDDs); Infrastructure (everything from elevators to smart meters to rail cars to PoS systems to smart city products, batteries, etc etc – maybe ~25-30% of group sales); and Energy (Nuclear and non-nuclear large scale projects, fuel & maintenance, etc – maybe 30-35% of group sales).

Due to the complexity I will not dwell too much on the Infrastructure segment (except Landis & Gyr) – it has historically been a collection of low margin, low cash return businesses and grows with some correlation to Japanese and global GDP growth. The business has historically taken its fair share of lumpy losses (eg the 70bn goodwill impairment in the PoS business this year) but Tosh is forecasting a return to low single digit (~3-3.5%) operating margins and low/mid-single digit topline growth. Those forecasts seem reasonable to me and frankly are very difficult to granularly build up to from the bottom, given the diversity of products within the segment. I believe it is more fruitful to look at the challenges facing some of Toshiba’s other segments.

Memory: Memory has long been Toshiba’s earnings pillar – for most of the past couple years (before a rapid deterioration in the last few quarters), NAND was a consistent operating profit generator (if not a massive cash generator), thanks to its favored cost structure (Japanese operations in Yen vs USD contract sales via its JV with Sandisk) and supportive NAND prices. Toshiba has never fully broken out NAND margins and earnings independent of its wider ‘Electronic Devices’ segment (which includes HDDs as well as what remains of LSI/Discretes), but my estimates suggest NAND alone was responsible for as much as ~75% of group operating profits in FY13 and more than 100% in FY14; by my math NAND profits peaked at ~240bn in FY15 when operating margins were around 27%.

The problem of course is what has happened since then, and more importantly, what will happen over the next couple of years. The back end of the planar, 2D NAND phase over the last couple years was a huge profit generator for all the main players (Samsung, Hynix, Micron and Toshiba/Sandisk) as a stable oligopoly enjoyed the fruits of the secular growth in smartphones, increasing SSD adoption at the expense of HDDs in servers and enterprise, and lower-than-normal capex burdens as planar NAND matured. But as in any cyclical industry, the above-average margins enjoyed for a couple of years only sowed the seeds of the current situation: a massive industry-wide ramp in capacity, a stagnating growth outlook as the smartphone adoption curve slows rapidly, and more price competition from increasingly less-disciplined players whom have seen their other profit centers (like DRAM and/or cellphones, for the likes of Samsung) rapidly diminish. In addition you have had the entrance to the market of a new, irrational, state-backed player (namely, Xinghua and the quasi-Chinese government), determined to build a strategic presence in the memory business with no regard for industry profitability in the interim. And compounding all this has been the inevitable technology transition, where the move to 3D NAND promises large cost per bit improvements on a multi-year view, but incrementally larger capex needs that depress profitability during ramp up.

To put a point on it: Toshiba is one of the biggest losers from the current situation in the NAND industry (perhaps Micron is the other big loser). The reasoning is fairly simple: Toshiba is the furthest behind the 3D NAND adoption curve (perhaps equally with Micron, perhaps a bit further behind), and at the same time is the most delayed in adding incremental capacity (all the major players – Samsung, Hynix, Micron, Intel, the Chinese, and now Toshiba – are adding new capacity thru the next 3 years). Thus – while Samsung (and Hynix to a lesser extent) had invested exponentially large $$ in R&D through the cycle and was mass producing 3D NAND even a year ago (Apr’15) with commercial shipments from at least 2H’15, the likes of Toshiba and Micron are now only beginning to mass produce their versions of 3D NAND, and hope to ramp to commercial shipments in 2H this year. In the meantime, Samsung et al are enjoying all the fruits of their cost advantage, having already ramped, and are thus in a position to happily push prices lower, gaining share while still making consistent margins at lower ASPs.

The likes of Toshiba, meanwhile are forced to accept lower ASPs whilst still laboring with the majority of their lines on planar production technologies – all the while consuming incrementally more capex dollars to try to catch up on 3D NAND. Is it any surprise, then, that Toshiba NAND margins look set to fall to ~5% in 4Q (Mar’16) this year and probably continue heading lower? Micron NAND operating margins have already turned negative and the recent quarterly report included some fairly dour comments on both supply and demand in the near-term (simply put, there is no end in sight to ASP pressure). The sad (for Toshiba) reality is that despite the secular shift to SSDs and the increasing NAND usage that this has caused and will continue to cause, being so far behind in the technology transition means most all the benefits of this shift will accrue to the bigger, better capitalized, R&D-heavier players who positioned much earlier than Tosh did for the shift.

This is where I have a big problem with Toshiba’s forward guidance. In the Mar 18th business presentation, they suggested NAND margins would be ‘at least 5%’ for FY16 (Mar’17), and then return to ‘above 10%’ in the next two years. This, however, is a massive assumption at fairly large odds with what is going on in the market and the recent behavior in the industry; it is also at odds with what we know tends to happen to commoditized, over-supplied markets with the entrance of irrational players, as well as being at odds with the specific history of the memory industry. In reality, I expect NAND average operating margins to continue to decline well into negative territory (as is currently the case for DRAM) and has been the historical norm for cyclical lows in the memory industry in the past.

Storage: Some do not realize, but Toshiba is still a significant player in the HDD market, with ~15% global share. The problems facing HDDs have been well analyzed by others and I do not mean to rehash them here; I simply bring this unit up as it is a great example of how Toshiba’s guidance remains disconnected from reality. Again, in the same Mar18th presentation, Toshiba suggested they will GROW HDD revenues 30% YoY in FY16 – this despite the fact that their HDD sales likely fell ~14% in FY15, and only the other day Seagate pre-announced horrible guidance on both top and bottom lines. More incredulous is the claim that this growth will be driven by new product offerings in the nearline cloud-focused segment – basically the only sub-segment of the HDD market that is growing (and, thus, increasingly competitive). In fact, Toshiba’s historical focus in HDDs has been on notebook/PC smaller sized offerings – the particular part of the market that is seeing the most rapid secular decline and that STX called out recently for being the most price-competitive.

Putting it all together, I would be shocked to see Toshiba maintain flat YoY sales in HDD at ~430bn JPY next year. Furthermore, Toshiba took some restructuring pain this fiscal year in HDDs (around 45bn) but given the ongoing scope of this business as well as the pace of secular decline as presaged by the better operators like STX, I expect that even break-even next year will be very tough to maintain. Thus I believe Tosh’s forecasts for both rapid growth and a return to profitability in the unit to be a nothing more than a pipedream, and we are likely to see further losses and asset impairments in this segment over the next year.

Nuclear Energy: Beyond memory/storage, Toshiba’s forecasts for their nuclear business are similarly devoid of any historical precedent or reasonable basis in fact. As part of the attempt to increase disclosure post the accounting scandal, Toshiba actually disclosed historical profitability for its nuclear unit (both the acquired Westinghouse, since 2006, as well as the legacy Toshiba nuclear business), in an attempt to ‘come clean.’ Unfortunately (for Tosh), these disclosures only made their go-forward forecasts seem more ridiculous. To highlight a few basic points:

-          Since 2006, EBITDA margins across all nuclear have been around 9% on average, with profitability peaking pre-Fukushima at 12% in FY10, while recent years have been in the 4-7% range

-          The construction business – as distinct from the fuel + maintenance segment – has only had three years of profitability in the last ten (and even those margins were 3-4%), and has been loss-making for the last four years straight

-          Absolute EBITDA peaked out at 73bn in FY10 and in the last two years has only been around 20bn

-          DESPITE ALL THIS, Tosh guides AVERAGE annual EBITDA for the ELEVEN YEAR PERIOD FY18-29 to be 180bn – or 2.5x peak EBITDA enjoyed in but one year, pre-Fukushima, of the last ten

Let’s look at some of the forecasts underlying this rather bullish set of earnings assumptions:

-          Company guidance: Toshiba will win 64 new reactor orders by 2029, out of about 400 globally planned for construction (ie ~16% market share). REALITY: Toshiba has not won a single new order in the last five years (since Fukushima), and as many 1/3 of the planned bids make no economic sense in the ‘new oil era’

-          Company guidance: Toshiba will generate an average of 60bn in construction EBITDA each year between 2018-2029. REALITY: Toshiba peak construction EBITDA was ~4bn (in 2010), and has been loss-making for the last four years

-          Company guidance: Toshiba will learn from its initial AP1000 (reactor design) experiences to achieve cost efficiencies and on time delivery for future orders. REALITY: existing projects for the AP1000 (for all but one project in China) are in developed countries (UK, US), and in any case are uniformly 2-3+ years behind schedule and over-budget. There is no historical evidence that Toshiba can deliver on time even in good jurisdictions

This should give you a taste of how ridiculous the company’s targets are. Toshiba also likes to point out that the goodwill impairment threshold is not actually 64 units, but only 46 – nevertheless this would still amount to an average of 4-5 new orders a year, for the next 8-9 years (given the lead time to construction by 2029), and again – Toshiba has received NO new orders since Fukushima and has not delivered a single project on time and on budget. In fact, Toshiba was basically forced to buy out CB&I’s nuclear construction business, Stone & Webster, to try to get the troubled Vogtle and VC Summer projects in Georgia and South Carolina back under some semblance of control and budget, given multi-billion dollar cost overruns and at least a 3yr delay to launch. Projects from China (Sanmen) to the UK (the Nugen JV at Moorside) face similar problems – indeed the Nugen project is likely to face funding issues in 2018 when a final investment decision is required, due to Toshiba’s new-found financial difficulties, the inability for Tosh to fund their share of the project, and the new (and dented) economics of nuclear power in the new era of low gas prices.

Indeed, it is worth focusing on the macro picture here for a moment. Pre-Fukushima and when oil was >$100/barrel, there were significant tailwinds for nuclear adoption, both in the developed and emerging world. But after the Japan disaster, much of the West (read – all of Western Europe, essentially) has abandoned nuclear power as a long-term energy source. Meanwhile the advent of cheap natural gas in the US has essentially neutered the nuclear rationale in Toshiba’s two home markets – Japan and the United States. Japan’s reactors will one day be operational again – though there will never be another new nuclear reactor built in Japan, mostly due to popular backlash against the industry, and indeed many extant reactors there will not live out their expected ~40yr lives (due to higher safety standards).  Meanwhile in the US, low gas prices make new nuclear projects – with all the financing, implementation, and execution risk they entail – essentially non-starters. Since ~65% of Westinghouse’s historically completed nuclear reactors are located in the US – as well as 4 of the 8 Westinghouse currently has under construction – this is not exactly a great outcome for the future of their nuclear construction and maintenance businesses.

Indeed this has already been demonstrated with the effective suspension of the South Texas Project for lack of financing last year (this cost Toshiba ~50-60bn in total impairments), as the project no longer makes economic sense with gas prices where they are. While Toshiba likely didn’t count too much on the US as a growth engine for new nuclear plants and while there are still growth opportunities in places like India and China (overwhelmingly the two major markets that all nuclear players pin their hopes on), the market has undeniably shrunk and thus the competition for new orders amongst the major players has only become more intense. The IAEA, as well, has recognized this new reality repeatedly over the last few years, steadily lowering its forecasts for global nuclear power generation, every year since the Fukushima accident in 2011.

This is where Toshiba’s junk rating and limited financing ability will become a real problem. As alluded to earlier, financing is a very important component in the bidding process for new nuclear construction projects: other than a nominal deposit, most all providers agree to extend financing until the completion and delivery of the project, in essence making any new project win hugely cash consumptive during the multi-year (even decade) construction period. This makes an investment grade rating, or otherwise the ability to raise large amounts of financing at competitive terms, incredibly central to the drive to win new business. But Toshiba is already suffering from its precarious fiscal position: the UK press reported that Toshiba had asked the Japanese banks to provide Tosh’s share of the financing for its Moorside project (part of the NuGen JV) – an extremely unusual move that the Japanese banks likely are balking at, and – in my view – will ultimately lead to Toshiba abandoning the project by 2018 (probably leading to more impairments). In any case, the upshot of Toshiba maintaining its current state of over-leverage is a necessary decline in forward business opportunities, which in turn will only hasten impairments in this business and thus a vicious cycle – this, again, is one of the main reasons why I foresee a large equity raising as soon as practicable.

Landis & Gyr: Landis & Gyr (L&G) is one of the less problematic areas of the Toshiba mélange, but here too there remains significant risk of earnings disappointment and intangible impairment over the next couple of years. L&G, a smart-meter production company, is, according to competitor Itron, a leader in electricity meters (with ~9% share), but is not really a major player in other types of meters such as gas and water meters. Nevertheless, Toshiba growth targets mandate fairly aggressive worldwide growth – again, despite the fact that earnings have been essentially flat since acquisition. Thus, in FY12 – the first full year of acquisition – L&G did 18bn in EBITDA; this year (Mar’16) Toshiba estimates it will do 20bn EBITDA on ~17% cumulative top line growth (over the last 3yrs). However, Tosh forecasts the company to increase sales and profitability by a step function in FY17, suggesting the top line will jump 20% and EBITDA will expand to 28bn (and then to 39bn in FY18). Given relatively flat depreciation, these numbers imply a doubling of operating profits (9bn ->17bn) in FY16->17, then another big jump to 28bn in FY18.

 Once again – within the context of Toshiba’s aggressive guidance elsewhere and much more modest historical performance, it is reasonable to be quite skeptical of this guidance. This is underlined by the more modest tone evident at some of Toshiba’s competitors (like Itron), where developed market meter penetration is already high; whilst in emerging markets, price competition is intense.

The cost of missing forecasts, is, once again, a large potential haircut to intangibles. Tosh has ~180bn in goodwill on balance sheet associated with L&G (~30% of pro forma book equity), along with ~40bn of other intangibles. Even a small miss to aggressive out-year forecasts could thus imply a meaningful hit to book equity in the form of a partial writedown of goodwill and/or related intangibles.

Pro-forma valuation is very expensive: The final piece of the puzzle is valuation. This is a little bit of a thorny subject given all the moving pieces so I gravitate to a multi-part approach: sum-of-the-parts, price/book, and price/normalized earnings (EPS). At the outset I should mention that price/normalized FCF is not really a usable metric given Toshiba historically generates very little free cash flow, and on a go forward basis I believe it is somewhere between chronically cash flow negative and break even. Of course, this ultimately only underlines the bear case as Toshiba has and will continue to generate horrible cash returns.

Method 1: sum-of-the-parts. Despite the fact that Toshiba will soon reorganize its segments, we will use legacy segments (ex healthcare) to determine fair value today use a comp multiple approach for each of the remaining pieces. My methodology for each segment was as follows:

-         Energy & Infrastructure: apply a 7x EBITDA multiple to pro-forma FY17E earnings. This is the high end of relevant Japanese comps which trade in the 4-7x range (Hitachi, Mitsubishi Elec, Sumitomo Heavy, IHI)

-         Community Solutions: apply a 7x EBITDA multiple to pro-forma FY17E earnings. This is the high end of relevant Japanese comps that trade in the 3-7x range (Toshiba TEC, Hitachi, Brother Industries, Mitsubishi Elec)

-       Electronic Devices: apply a 6x EBITDA multiple to pro-forma FY17E earnings. This is many turns above the listed US/Korean memory multiples (MU trades at <3x, Hynix at <3x, WDC at ~3.5x, STX at ~5.5x) while SNDK was acquired at ~9x with a very large acquisition premium

-       Others: apply a 6x EBITDA multiple to pro-forma FY17E earnings (small segment, all the other hodge podge)

-          In addition – I give Toshiba full credit for all asset sales proceeds announced, treat securities investments held on balance sheet as cash (~123bn), treat the unfunded pension as debt (689bn), treat off-balance sheet debt guarantees as debt (125bn, excluding LNG purchase commitments)

-          Also, I do NOT penalize Toshiba with any conglomerate discount (though other Japanese conglomerates, like Softbank/Hitachi/trading companies, generally do incorporate an additional 10+% haircut for this)

-          This generates an implied equity value of ~80 JPY/share (vs 230 JPY/share currently, or 65% downside). The implied composite EV/EBITDA is also 6.6x, which looks rich versus where the closest comp – Hitachi – trades, which is itself a much better collection of businesses with less leverage/balance sheet issues and no problematic memory exposure

Method 2: Price/book. Given the variability of earnings and chronic low capital returns, many equity analysts simply try to use pro-forma book value as a floor. I don’t really agree with this approach – especially since, as discussed, tangible book has been and remains strongly negative, so I don’t think that is a floor. Indeed, most Japanese comps (again, Hitachi, but also Mitsubishi Elec, IHI, etc) trade at reasonable discounts to book value. In any case – pro-forma book value for Toshiba – assuming NO equity dilution and NO further impairments to intangibles – is ~146 JPY/share (~620bn in pro-forma equity dividend by 4.238bn shares out). This alone is 37% downside to current, and again – I think there are still large book impairments to come from a variety of sources. For example, simply haircutting book for the likely 200bn impairment to Westinghouse intangibles gets you to ~100 JPY/share.

Method 3: Pro-forma multiple on ‘normalized’ EPS, post expected dilution. This is what I view as the most accurate way to value the stock today (and, I believe, is where the street will gravitate towards over time). That is – we estimate the normalized earnings picture and then adjust the share-count for how much new equity we believe will be issued by Tosh to delever the balance sheet. As mentioned earlier – I believe Toshiba needs somewhere between 300 and 400bn in new equity to get somewhat comfortable. If we assume they could sell, say 350bn of new stock at 190 JPY/share (this may be aggressive), that would imply 1.84bn new shares out or ~42% dilution versus current share count (~4.24bn). My current net income estimate for FY17 is fairly close to my normalized, go-forward estimates (that is, I expect memory +nuclear earnings to not recover in the medium term, off-setting any incremental improvement in the industrial/infrastructure part of the business). Since I currently expect Toshiba to post ~47bn JPY net income next year, this would be ~7.5-8 JPY/share on an adjusted new share count of ~6.06bn shares.

It is worth keeping in mind that EVEN in this scenario, Toshiba would still carry pro-forma adjusted net leverage of ~4.8x EBITDA (1340bn adj net debt vs ~280bn in EBITDA) so you could still argue this is far too levered. But in any case – even at ~8 JPY/share, given where the best comps trade, and the underlying poor economics of most all of the businesses, a multiple north of 10x would be a stretch (Hitachi, for example, trades at ~7x FY18 consensus EPS). Hence, I envisage a similar target price to that implied by the SoTP, or ~75-80 JPY/share, or perhaps lower (thus again, 65%+ downside).

To try to quantify the downside to the trade – Toshiba has guided to ‘at least’ 100bn in net profit in FY18 (ie, 3 years out), which is itself based upon more than doubling operating profits from ~120bn to 270bn over the next 3 years. Clearly I do not believe Toshiba’s medium-term forecasts for one iota (they are predicated, as discussed, on a massive recovery in nuclear and particularly memory), but EVEN IF they came true, on our new diluted share count of 6.06bn shares, this would imply EPS of ~16.5 JPY – which, at Hitachi-like normalized multiples (say 8-9x), in turn only suggests a stock price 132-149 JPY – ie, still at least 35% down from here, with all the negative optionality not even thrown in.

A couple of additional points:

-          There is no value to the equity in the SoTP if you give the Electronic Devices business a market multiple. That is to say, if the market valued Toshiba’s middle-of-the-pack NAND business, and worst-in-class HDD businesses, at where its US comps trade (somewhere between 3-4x EBITDA), Toshiba stock would essentially be a zero

-          Bulls may argue that my implied valuation (~6.5x EBITDA on SoTP and ~high single digits PE) are meaningful discounts to where the stock has historically traded (7-8x EBITDA and mid-teens PE the last 5-6yrs). This may be true – but I would counter that Toshiba deservers to trade at all-time troughs in terms of valuation given the deterioration in balance sheet quality, chronic over-leverage, dire forward outlook in a number of core businesses, and – frankly – the inability of the stock historically to capture the low quality of the underlying businesses (as was made clear by the accounting restatements)

-          Current valuations look extremely stretched: at 230, the stock is trading at ~9.5x EBITDA, 1.5x book, and ~21x EPS – assuming NO dilution – based on my FY17E numbers. This is insane versus direct memory and storage comps (as discussed), as well as direct Japanese industrial comps (generally <6x EBITDA, below book, and low teens or single digit PEs). I can only explain this by referencing the large short position in the stock, which was a hangover from a different trade (the accounting/restatement over most of last year) and some ‘fatigue’ with the story as near-term, direct catalysts appear to have mostly played out. The story from here has evolved into one of chronic business underperformance with the still not-insignificant risk of further balance sheet impairment.


Risks/what could go wrong: Other than management outperforming, let along hitting, their medium-term targets, the biggest risk is some kind of further asset sale at an irrational price (as we saw with the Medical business). While this was one of the proximate causes for the relief rally, given Tosh has little left to sell (other than stockholdings, which would not generate book equity gains needed to delever, just cash), this is much less of a risk going forward. The actual risk the entire entity gets sold, or even a piece of one of its main businesses (nuclear or NAND) is very small. In nuclear – any sale would necessarily catalyze a large impairment, given where valuations are today versus a decade ago: this is why the company has chosen not to sell down a minority stake in the last couple of years (as they admittedly tried to do). Meanwhile there is simply no viable buyer for even a piece of the NAND business, now that WDC has bought out Sandisk – effectively giving them veto power on any potential sale of the remaining piece (ie, Toshiba’s) to another competitor. Even if we assumed WDC would allow a competitor to buy a minority piece of Toshiba’s stake, there are relatively few buyers that would make sense, outside of Chinese competitors (since Seagate is likely already too levered to contemplate that kind of acquisition). Thinking imaginatively, perhaps a Japanese quasi-governmental entity like the INCJ would pony up for a minority stake – but even then I don’t think that would happen unless the entire Toshiba group became much more distressed. It is all rather a moot point, though, as Toshiba has strategically doubled down on memory by committing to large scale capex to build out another fab at Yokkaichi, meaning they are likely in memory for the long-term, for better or worse. Meanwhile, a sale of the HDD business would be putatively very difficult, given the declining nature of the business and already-oligopolistic nature of the industry.

The other main risk as I see it is simply what has led to outperformance the last month or so: somewhat crowded positioning on the short side, and general fatigue with the bear case. I believe this will change over time as the market gradually realizes the ‘impossible dream’ nature of management’s recovery targets; and as large equity dilution becomes more imminent. Also, the already-extreme nature of pro-forma valuation vis-à-vis comps gives me comfort for a medium-term holding period. That said – the recent rally has been painful and there is no guarantee the stock cannot keep squeezing for a while.

I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise hold a material investment in the issuer's securities.


- Quarterly earnings releases
- Sudden intangible impairments
- Secondary equity offering

    show   sort by    
      Back to top