Description

GATX is a highly leveraged specialty finance business that the market has mistaken for a “late-cycle” rail company. We argue that the widespread distribution and subsequent commoditization of railcar leases as investment vehicles combined with QE has permanently lowered asset yields, resulting in low ROEs as GATX is now one of the high cost players in the space versus deposit-funded competitors. While the Street expects gradually rising earnings to ~5.60 per share in several years we see the potential for $2.00 or less over time, implying downside of 60%+

Thesis:

• Secular margin compression in the largest operating segment (Rail North America) due to the commoditization of railcar leasing and entrance of competitors with lower cost of funds
• Widespread adoption of precision scheduled railroading implies a step-change downward in the quantum of railcars needed in the North American fleet, disproportionately impacting older railcars owned by GATX
• “Gain on sale” income the last 5 years (25% of pre-tax profit) is unsustainable.  Gains have been due to falling asset yields referenced in the point above. As above-market coupons naturally reprice to market rates on lease expiries, the 45-50mm of rail North American gains on sale should diminish substantially
• Expiry of the “super profit” leases from 2013-2014

The stock has episodically been a popular short for the last several years on thesis of the unwinding of the 2013-2014 crude-by-rail boom surplus profits. GATX was written up as a short a little under 2 years ago by Pop4Pres. We agree with the thesis, albeit view the timing as being in two parts: in 2020 and 2024/2025, given our analysis of super profit lease expiry schedule. We view the RoE (and EPS) compression as inevitable given sharply falling reinvestment returns, unsustainable gain on sale income in recent years, and the expiry of 2013-2015 leases. 2020 numbers should begin to tell the story.

Rail stock or nonbank financial?

Equity analysts who cover the stock primarily cover the rail industry. As a result, they appear to be singularly fixated on the one readily available industry statistic that drives revenues for the Class 1 rails – the volume of carloadings – rather than the underlying factors that drive railcar leasing rates – asset yields, funding costs and required returns on equity for leasing companies. As a result of their orientation to volume statistics, analysts appear to take GATX’s suggestion that current conditions are cyclically “weak” at face value in order to apply a higher multiple on supposedly early cycle earnings. Indeed, GATX’s current P/E multiple at 16x FY1 consensus estimates is significantly closer to trough multiples of ~20x (in 2010) than peak multiples of ~8x (in 2015) 

Source: GATX presentation from 11-13-19

GATX’s slide begs the question – what cycle are we talking about exactly? Analysts generally seem to think that from a macro perspective, the company’s earnings are driven off the US manufacturing cycle, and hence the relevant data are US manufacturing PMIs and international trade volumes. This is understandable, given the importance of these metrics historically to US rail volumes and to Class 1 rail stocks. The theory goes that as US manufacturing picks up, US rail loadings will increase as well, eventually translating into more demand for both owned and leased railcars, driving GATX revenue higher.  Given widespread investor optimism about a US (and to a lesser degree, Chinese) manufacturing recovery in 2020 on the back of improved trade dynamics, the market bid up the shares of GATX in fall 2019 along with everything else perceived to be cyclical.

Our issue with the argument above is that it presumes that GATX still operates in the industry structure that existed 15 years ago, not the one that exists today. The issue is that the variables needed to forecast demand and revenue for Class 1 rails and railcar OEMs are only a subset of those which determine railcar lease rates. The drivers for the total number of active railcars in the North American fleet are 1) the amount of carloadings for each commodity moved by rail and 2) the efficiency of the class 1 rails in moving those carloads, which is proxied by train velocity and dwell times. The demand function for the volume of new cars made each year is a function of changes in the total fleet required and the replacement rate of old cars (scrappage). The demand for the volume of leased railcars is a function of shippers’ preference on owning versus leasing rail equipment and their financial ability to own cars. Because railroad efficiency, scrappage rates, and shipper preference are less visible variables, the market has tended to hold them constant and assume that higher carloadings translates directly into higher railcar lease volumes.

However, GATX does not earn revenue based on the total number of new cars produced or carloadings but rather on the lease rates for rail equipment. The primary drivers of leasing rates in the last few years have not arisen from the demand side of the equation, but rather from the behavior of the railcar OEMs that control the supply of available railcars. The creation of railcar leasing & syndication subsidiaries by the two dominant manufacturers (Trinity and Greenbrier) in the last 10 years permanently impaired the profitability of GATX’s core North American leasing business by making lease supply a function of asset yields (e.g. fixed income demand) and depository required returns on equity. Railcar leasing used to be a niche industry where the supply of railcars was determined solely by end-user demand. Widespread syndication of railcar leases transformed the leases from an end-user product to a financial investment product. In a world of global QE, the amount of demand for yielding assets has allowed the supply of low yielding railcar leases to proliferate well in excess of that required by an analysis of railcar loadings.

For some quick background, the railcar manufacturing business has been terrible over time. At the risk of gross simplification, the creation of a traditional freight car basically involves bending steel and welding, featuring tight gross margins (low teens percentages on average) and high fixed costs. Historically this led to feast & famine dynamics at the OEMs who printed money in economic expansions with rising demand and hemorrhaged losses in recessions when demand declined. GATX was seemingly the only cyclically-aware company in the industry, locking in long multi-year lease terms during expansions to provide stable revenues in recessions. The other players chased booms and inevitably had to sell out at depressed prices a few years later as they raised cash to survive.

After 2008 the OEMs realized that they needed a more stable profit stream and significantly expanded their captive leasing subsidiaries (similar to US auto OEMs decades prior) and railcar investment platforms. In essence the OEMs transformed a traditional railcar lease from an esoteric service provided to shippers to a tradeable, commoditized, fixed income asset. The OEMs provide leasing & maintenance services, allowing third party balance sheets to access railcar lease income on a wholesale basis. This development worked for the OEMs because it provided a steady stream of financial buyers for their railcars, smoothing out the manufacturing volatility and keeping the factories utilized, but it destroyed GATX’s ability to earn surplus returns forever. The global reach for yield that accompanied quantitative easing the last 7 years greatly furthered this dynamic by bringing in banks with 0% to 1% deposit cost of funds to compete with GATX as owners of US railcar leases. In essence GATX went from being one of the best capitalized players in a niche market to a high cost of funds player competing for yield in the global capital markets.

GATX is not a rail company. It is a non-bank financial services company whose customers are in the rail industry. Given the widespread syndication of railcar leases GATX’s returns are now a function of global interest rates and the RoEs required by its US and Japanese bank competitors who have a 200-300bps cost of funds advantage despite the BBB rated unsecured market trading at record tight spreads (where GATX funds). If Trinity and Greenbrier are the Ford and GM of the railcar industry, GATX is the Ally Financial, minus the deposit funding. While GATX says that they don’t think about their business in terms of lease rate factors, everyone else in the industry does. Simplistically the lease rate factor is the monthly lease payment divided by the cost of the leased equipment. A new large cube covered hopper that serves the plastic pellet industry cost ~$100,000 as of October 2019. The monthly payment for a full-service lease was around $500, or a lease rate factor of 0.5x. Multiply the lease rate factor by 12 and you get the annualized asset yield, same as a bond (6.0%).

To get a sense for the scale of asset yield compression that has occurred in the railcar leasing industry, in the 1990s the lease rate factor was ~1.0. In the 2000s credit spreads compressed and the lease rate factor dropped to 0.75x (9% asset yields). In 2012 post QE and the commoditization of rail leasing it dropped to 0.5x on average (6%) and recently has been as low as 0.45x on new freight car railcar purchases. Generically speaking freight cars leases have been 0.5x, while tank car leases have been 0.6x to 0.65x. While GATX would say that these asset yields are insane and unsustainable, consider that these are effectively secured credit exposures to predominantly investment grade rail and shipping customers, and the average USD unsecured bond yield is 3%. Said differently, the yields are not unsustainable for a deposit funded institution, only for GATX.

For illustrative purposes, here are some numbers as of early fall 2019 on new car investments to show the impact of QE on the market. Although there are no regularly published price statistics for leases or new cars, conversations with industry participants can easily confirm market prices and dynamics. The reason we use early fall 2019 figures is that the market has deteriorated significantly since then (both lease rates and new car prices down in $ terms) and is hard to pin exactly today because things are in flux (https://www.wsj.com/articles/banks-own-thousands-of-railcars-but-dont-know-what-to-do-with-them-11577356201).

• Freight car leases averaged ~$350-400/month, ranging from $250 for a small cube covered hopper (frac sand) to $500 for a large cube covered hopper. Tank car leases averaged $700/month, ranging from $350 for a legacy DOT111 to $1000/month for a new DOT117 crude-by-rail car. Small cube covered hoppers cost ~$75k (4% asset yields), large cube ~$100k (6.0% yields), and tank cars $150-160k (5.5%-8.0% yields)
• Railcars traditionally have 40 year useful lives and are depreciated to 10% residual value (scrap value). This means a 2.25% real depreciation charge per year (90% of asset value / 40 years). Some companies present cash-on-cash yields that back out depreciation but given the wasting nature of the asset it is a real expense from an economic perspective – otherwise return of capital becomes confused with return on capital.
•  $65-$200/month in maintenance accruals (65 for freight, 200 for tank), $20 admin, management, and insurance costs, $10 ad valorem taxes for 95/month operating costs for freight and $230/month for tank cars. Maintenance statistics are widely variable for tank cars. There are very little costs incurred in the first decade of a tank car’s life, but they increase dramatically later in life due to the need for very expensive regulatory tank car qualifications, so generalizations are difficult given varying railcar ages and the complexity of a given tank car (e.g. coiled and insulated tanks cost more to maintain than a general service tank car). Given the leasing industry’s recent practice of sticking customers with huge repair bills upon car returns at lease expiry (i.e. GATX ‘other revenue’), customers naturally would prefer to lease a newer car rather than an older one at the same rates to avoid an unpleasant surprise.

Worked investment examples:  We chose the following two car types because plastic pellet cars (large cube covered hoppers) have been the “it” car for the last few years on the freight side, attracting the most investment demand, while crude by rail cars have been the same on the tank side. We would argue that these numbers are indicative returns on new investments in a QE world, while GATX reported financials currently benefit from back-book leases (particularly the super profit leases signed in 2013-2014) and gain on sale, neither of which we think are sustainable longer term.

• Plastic pellet cars: $100k ASP, 6000 annual revenue, 2250 economic depreciation (2.25%), 1140 operating cost ($95/month) = 2610 annual profit prior to funding costs. The nonbank leasing industry operates with 75% debt to capital, so a US bank with 1% deposit funding costs would have $750 interest, or $1860 pre-tax returns, a 7.4% RoE on 25k. We use pre-tax RoE because accelerated depreciation means the leasing subsidiaries essentially never pay cash taxes as long as they are reinvesting into new railcar assets. GATX has a 3% cost of funds, or $2250 on the $75000 of debt, making this lease slightly profitable for them ($360 of annual profit, 1.4% RoE). In reality the bank probably uses a higher implicit leverage (8x, 12.5% CET1 ratio) if they are funding the purchase out of their regulated subsidiary instead of a holding company, getting them to ~14% returns. The economic depreciation in this case is a real charge since that cash flow would be needed to reinvest in new lease capex to keep to keep the fleet size constant and avoid paying taxes or else pay down the principal of the debt if liquidating.
• Crude by rail tank car: $160k ASP, 12000 annual revenue, 3600 economic depreciation (2.25%), 2760 operating costs (230/month) = 5640 unlevered profit, or 4440 profit to a bank (11% RoE) after 1200 in funding costs. GATX incurs 3600 interest costs, yielding a 2040 pre-tax profit or 5% RoE.

If the front-book RoE calculations from above seem low, we can compare them to Trinity Leasing’s own disclosure from March 2019 which shows their book yielding a 4% blended RoE in 2018. (The cash-on-cash figures add back depreciation and include revenue from sales of railcars owned less than 1 year, and hence is not comparable to GATX RoEs).

Source: Trinity rail March 2019 presentation

The main exception to the supply/demand/asset yield discussion presented above is when speculative demand for a given car type temporarily outstrips OEM manufacturing capacity, thereby spilling over into demand for used car leases. This was most prominent in 2013-2014 when record Bakken crude-by-rail shipping profits led to parabolic demand for new tank cars, overpowering OEM tank manufacturing capacity and leading to multiyear backlogs. As the OEMs were not able to meet demand, lease rates on used tank cars were bid up to extreme levels (and extreme lease lengths) as detailed at length in Pop4Pres’ writeup. This dynamic occurred very briefly again in late 2018 and early 2019 when investors placed orders for thousands of crude by rail tank cars destined for Canada. These orders took up all available manufacturing slots for 15-17 months in late 2018, with the first new tank cars available for delivery in April 2020. The temporary shortage of new car manufacturing capacity tightened the used tank car spot lease rate market, but this has rolled over again as follow-on car orders did not materialize. The first available car delivery slots still remain April 2020, allowing shippers to be selective and take advantage of competitive lease rates on new cars from OEM leasing subsidiaries rather than the used car market. OEMs can offer shippers the option of leasing a brand new railcar at the same or cheaper price as the expiring lease on their old cars, ensuring that the new car gets built at the expense of a subsidized lease rate and lower used railcar residual values. As the OEMs generally have new railcar fleets owned and under management, the primary losers in this dynamic are the owners of older railcars – namely GATX who has an average fleet age of 20 years.

Precision scheduled railroading implies a step function downward in the rail car fleet online

While the market has gotten overly excited about a US manufacturing recovery stemming from a trade deal, which would drive carloadings higher, it seems to have overlooked the impact of precision scheduled railroading dramatically increasing railroad efficiency. PSR is a term used to denote railroad efficiency programs that started with Canadian Pacific in 2012 and have subsequently spread to a number of large Class 1 railroads – UNP, NSC, CSX, and KSU are implementing PSR, with BNSF periodically mentioned as pursuing different efficiency programs not called PSR. According to the Class 1 narrative, railroads used to focus on optimizing train length (the idea was as long as possible) to maximize capacity, but the result was a given train service episodically being canceled on days when demand wasn’t sufficient to meet minimum train length. PSR shifted the focus to moving cars rather than trains, such that trains are always moving and cars are picked up on a schedule regardless of train length. The outcome of PSR has been to reduce network congestion and dwell time, allowing cars to be more efficiently utilized. The more efficient utilization of cars means less railcars are required in general to move the same volume of freight, leading to a step-change in leased/owned railcar demand downward as railroads implement these efficiency initiatives.

As of end 2018 the four railroads definitely implementing PSR had ~675k railcars, or 40% of the North American fleet. Buckingham research on 9-6-19 estimated that PSR from those railroads represented a headwind to new car demand of 75k railcars (about two years of replacement demand for the industry), in addition to any fleet efficiency savings achieved by BNSF. We argue that the new cars will be built regardless and put on lease at competitive rates, pushing older cars into storage. During its first year of PSR implementation in 2012, Canadian Pacific achieved a 26% improvement in carloadings per average car online versus the prior year. As railroads grow more efficient, they need less cars online to handle the same volume of loadings, leading to more cars in storage assuming carloading volume is flat. The reality in 2019 is that PSR and weaker carloadings added 120k cars to storage over the course of the year.  We would argue that PSR and rail market share loss to trucking for intermodal volumes were responsible for the majority of the carloading decline in 2019, not “the trade war”, but it’s impossible to prove which blade of the scissor cut the paper. We do assert that continued PSR implementation should be more than sufficient to offset improved demand from carloadings going forward in terms of the number of railcars required online.  

In all honesty, the industry has a long way to go simply to recapture prior levels of efficiency from the last decade. The table below shows the average US loadings per car, and a reversion to historical efficiency levels would greatly increase the amount of cars in storage. Higher levels of cars in storage depress lease rates, as they compete with new car leases and existing lease renewals. At December 2019 cars in storage were 396k. This amount of storage was last seen in 2016, when spot lease rates were 20-30% lower on tank cars (~$500/month), implying a tougher spot lease rate outlook for 2020. 

Gain on sale income in Rail North America:

Gain on sale income at GATX has been presented as a naturally occurring phenomenon, which should continue in perpetuity at a $45-50mm annual rate. Railcars are simply 40 year fixed income assets whose coupon changes approximately every 5 years, assuming the car stays leased for its entire useful life – which is not necessarily the case for cars facing changes in regulation (flammable service tanks) or serving commodities in secular decline (coal hoppers). We would argue that the vast majority of gains on sale achieved by GATX since 2014 have been due to new financial investors extrapolating a high coupon (e.g. the existing monthly lease rate) for the car’s useful life in their IRR models. Hypothetically a plastic pellet car with $500/month lease rates and $95/month opex with 20 years of life remaining to the scrap value would be worth ~$59k at a 6% IRR. At $600/month on the same assumptions the car is worth ~$72.8k. In 2013 through early 2015 the lease rates on all car types were elevated due to a persistent shortage of manufacturing slots due to the crude by rail boom, creating vintages of “high coupon” lease rates. In 2016-2018 GATX has been able to sell the high-coupon monthly lease rates to buyers who assumed that high lease rate would continue for the remainder of the cars’ useful life, generally because of optimistic assumptions on carloadings for the relevant commodity. In reality, lease rates since 2014 have come back down as manufacturing slots opened up and new car supply met the existing demand. We argue that as GATX’s back book of 2013-2014 leases reprice to current conditions, gain on sale should diminish significantly, if not entirely, as a number of car types are actually underwater (loss on sale / impairment).

An additional headwind to gain on sale income is that fact that used car lease rates on ALL car types have declined in the last three months due to a persistent excess supply of railcars and the impact of PSR on reducing the number of cars needed online. We understand that both freight and tank car lease rates have dropped by 20% year-year in 2H’19 ($80-100/month declines for freight cars and $150-175/month for tank cars). Given the buyers’ habit of anchoring on current rates into the future on a 20 year asset, these drops imply materially lower sale prices in the used car market. GATX implied in their 3Q call that some gains on sale are being deferred from Q4’19 into 2020 to help with reported EPS (i.e. ‘timing of sales happened to get pushed into 2020 from Q4’), but even so we would think that overall gains should begin to trend lower going forward.

Super profit leases from 2013-2014:

This sub-analysis relies on a decent amount of guess work based on GATX presentations, but I include it because of the prevalence of the original GATX short case regarding crude-by-rail (CBR) “super-profit” leases from 2013-2015 to help on the topic of timing. The disclaimer is simply that the precise numbers which follow are almost certainly wrong given a lack of company disclosure, but the overall point is broadly correct.

The crude-by-rail boom took off in 2013 and began turning downward in the summer of 2015. We know that GATX signed a number of very long dated, super-profit leases starting in 2013 through early 2015 on flammable service tank cars that were capable of being used for crude by rail (DOT 111 and CPC 1232 tank cars). From conversations with industry participants, it seems that the longest leases done in the industry during that time were for 10-year terms. Spot / shorter-term lease rates were ~$2000/month at the peak in 2014 and term lease deals were being done at $1200-1400/month. From company disclosure, less than 2500 of GATX’s tank cars were in crude service at the peak. However, this figure is irrelevant because any tank car that could have been used for crude service had lease pricing based on crude service rates; crude-by-rail determined the lease pricing for all flammable service tank cars even if they were not being used in CBR service (e.g. carrying refined products, etc). Hence GATX received the benefit of peak crude-by-rail lease rates on all of the flammable service tank cars that it had in the fleet at that time.

Source: GATX presentations from early 2016 (December 2015 committed lease expiry schedule)

GATX gives a snapshot of their lease expiry schedule by year in $mm amount once a year, with the “thereafter” bucket the most interesting for our analysis. The above table is the lease expiry schedule as of 12/31/15 from company presentations (entire company, not just Rail North America), a point in time where committed lease receipts would include all super profit leases.  The $990mm of cumulative, committed lease receipts in the “thereafter” bucket (expiring in 2021 and later) at December 2015 is a truly enormous number that the company has never seen either before or after the CBR peak. The bold and boxed numbers below show how that bucket has trended since then.

Source: company presentations

Using the assumption that no leases were done in 2015 longer than 10 years and that in 2016-2018 there were very limited greater-than-5-year term lease deals on flammable service tank cars, it would appear that GATX locked in $200mm+ of annual revenue from peak rate leases during late 2013 to early 2015 which are largely still on the books today. For context that would represent around 13,700 tank cars at monthly rates of $1300/month. With the benefit of three years of hindsight, we can see that the cumulative “thereafter” number of committed lease receipts has been decreasing at a rate of $215-222mm per year since the 2015 peak. For example, in the year between 12/31/15 and 12/31/16 the amount of cumulative lease revenue expiring in the thereafter bucket decreased by $222mm. If we assume that zero greater than 5 year leases were done in 2016 which would have added to the “thereafter” bucket at 12/31/16 – a reasonable assumption given conditions that year and the fact that GATX’s new leases in 2016 averaged ~31-32 months of term – then 2021 committed lease revenue would have been $222mm as of 12/31/15 by implication. On that basis a possible disaggregation of the 990mm “thereafter” figure at 12/30/15 by year could be as follows:

Following from the table above we can back into an implied “super profit” contract expiry schedule by year from the end of 2015 in the table below:

The immediate observation is that a substantial portfolio of 2015 and prior leases expire in 2020, causing the lease repricing headwind to be twice as bad for GATX in 2020 as it was in 2018 or 2019. While the table represents committed lease receipts for the whole company, given that Rail North America was 66% of company-wide revenue in 2015 it’s safe to assume 2020 has a lot of North America super profit leases expiring. The other observation is that there appear to be a large number of very long dated contracts from that period still on the books, leading a second repricing to occur again in 2025.

Let’s assume for a moment that the $214mm of lease revenue in 2024 from 2015 and prior vintage leases was done at $1300/month, in line with industry conversations, implying 5600 cars coming off lease in 2024 and 8075 cars coming off lease in 2025.  Our understanding based on past management commentary in various investor presentations is that these cars were “legacy” tank cars such as CPC 1232s, if for no other reason than the current regulatory-compliant DOT 117 car wasn’t being made back then. Given changes in the tank car regulatory landscape in the last 5 years, CPC 1232s face significant retrofit costs for them to reach current compliance standards. 1232s require a $40k retrofit per car for non-jacketed cars and a $3k retrofit for jacketed cars in order become DOT 117Rs – which have been leasing at $500/month because some railroads will not accept the retrofit versions of the 117 at all and others charge excess tariffs on them.

A reasonable estimate would be the 1232 tanks getting repriced to $500/month in revenue, which would reduce operating profits by 95% considering that operating costs would be unchanged. This represents a potential aggregate reduction of ~$130mm in annual operating profit for Rail North America versus the ~$300mm realized in 2018. The reality is that the 1232 non-jacketed cars will simply be stored or scrapped when they come off lease, while future lease rates in 2024 will determine what happens to the jacketed 1232s. Regardless, the thesis that GATX has been earning abnormally high profits on 2013-2015 vintage tank car leases appears to be correct, but will ultimately play out in two phases – a significant one 2020 and a bigger one in 2024/2025. Management themselves noted at the Stephen’s conference in November 2019 that most of GATX’s tank cars currently in CBR service were still on their peak vintage lease. At the very least I would think that this lease expiry cliff would have some bearing on the P/E multiple that investors are paying for the company, since it is still clearly over-earning. 

Valuation:

Simplistically we use a 15x multiple on normalized earnings of <$2.00 per share for a $30 price target, representing a 0.6x multiple on 2021 tangible book value of $50 for a 4% ROTCE. The ultimate downside on GATX earnings will be determined by whether or not steadily rising leverage levels ever forces GATX to sell cars at a loss. A number of car types in their fleet would sell at a loss to book value today (open top and small cube hoppers) but GATX has avoided doing so to maintain its valuation at 1.7x book value. For the company to trade at the 0.4x TBV implied by forward returns a recession is likely required given eternal investor optimism on this stock. We note that most pure railcar fleet sales in the last 18 months occurred at significant discounts to book value (Element Capital, Ferrellgas), the implied valuation of CIT’s rail business is at a significant discount to book, and Wells Fargo Rail purportedly couldn’t be sold anywhere near book (which is why the sale was cancelled).  CAI, which has a young rail fleet, still had to impair the portfolio twice in 2019 to 90% of book value as part of preparing the portfolio for sale.

Part of the reason for portfolio losses on sale (particularly for older portfolios) is that the deferred tax liability becomes due in a change of control, turning into de facto “debt” in the enterprise value calculation and reducing the equity value paid to the seller. GATX’s DTL is $910mm ($25/share) meaning any “takeout” valuation is materially lower than the going concern value where GATX continues investing into new railcars to avoid paying taxes, even at diminishing returns. As a result of the materially lower return on invested capital on new railcar purchases versus historical investments, GATX has been free cash flow negative while attempting to maintain flat earnings despite paying $0 cash taxes, thereby increasing leverage.

Given the negative free cash flow to date in 2019, the company is also having to borrow to fund its dividends and share repurchases. GATX has also sought to protect earnings by shortening the maturity profile of its debt (incurring more commercial paper rather than term borrowing) and running down cash balances. As GATX needs to maintain its investment grade rating status (BBB rated) as a core part of its business model, we believe that the company is reaching its limitations on increasing leverage to maintain $ of net income. Therefore, we think <10% consolidated ROEs and a below book value multiple are defensible price targets because the alternative (a portfolio sale that accelerates the DTL repayment) is significantly worse. 

As a side note, people argue that GATX traded at 1x book value in 2009, so why should it trade for less than that in our price target? The answer is because the industry structure is materially worse than in 2009 and global interest rates (and required return hurdles due to QE) are much lower – not because we think that the economy will be worse. The fundamental argument is that interest rates and the funding costs of GATX’s competitors, not manufacturing PMIs or “the economy”, are what drive lease rates and hence EPS.

The forecast for $2.00 of earnings or less assumes simply that back book investment returns eventually converge with where new car investment returns have been for a few years. The forecast does not assume a recession (ever) as earnings would be significantly lower in a recession, and negative if GATX themselves were ever in a situation where they needed to delever by selling cars. The degradation in earnings is simply a function of QE and financial investors with lower cost of funding competing against GATX. We would argue that US regional banks and Japanese banks have a significantly stronger financial profile than GATX, so we question the common bull argument that they are the “weak hands” who have to sell cars at distressed prices to GATX – as if GATX’s fleet would somehow be the only highly performing cars in a recessionary environment.

The timing on the earnings declines will be governed by how quickly gain on sale declines, the impact of 2020 lease repricing, and the ultimate expiry in 2024/2025 of the super profit leases. The whole company is funded out of GATX Corporation, domiciled in the US for tax purposes. Our forecast envisions Rail North American profits declining substantially, which could trigger a limitation on interest tax deductibility in the US (interest expense > 30% of EBITDA is not tax deductible per the 2017 Tax Cut & Jobs Act), causing the effective tax rate to increase since all global interest expense is incurred in the US. Due to falling gain on sale income and 2020 lease expiries, we model EPS to be in the $3.50 context by the end of 2021 before dropping to <2.00 in 2025.

Risks:

Rating agencies permit a massive increase in leverage without downgrades: At the core, GATX’s problem is one of falling reinvestment rates of return – the front book simply generates less income per dollar of invested capital than the back book. To maintain constant dollars of profit, GATX has been increasing its investment into railcars beyond its free cash flow, taking on leverage in the process and increasing debt / leasing EBITDA ratios (e.g. operating profit excluding gain on sale). We argue that this increase in leverage will become more obvious as gain on sale income declines, and that rating agencies will not allow GATX to increase leverage ratios ad infinitum. It’s possible that rating agencies will give them a bit more slack in terms of increase leverage, but we would think that after 2020 lease repricing lowers income the company will be constrained.

Repair income surprises to the upside in 2020: Railcar leases contain conditions that require the lessee to return the car at lease expiry in the same condition as they received it. In 2019 GATX began aggressively enforcing this clause to bill the lessee for repairs when they return the car at lease expiry, which is recognized as “other revenue”. Other revenue increased by ~$25mm in 2019 year-on-year, a huge amount. Due to the high number of 2013-2015 vintage leases expiring in 2020, it’s possible that a number of them were damaged over the last few years, causing GATX to get a bump in repair revenue if the lessees choose to return the cars at lease expiry rather than release them at lower rates. In such an event GATX would get a one-time bump in repair income, but then have to find a new lessee or face an unutilized car, depressing earnings in 2021 and on. All things being equal, we think the aggressive enforcement of repair clauses over time will cause lessees to choose new cars from the OEM’s leasing subsidiaries rather than lease GATX’s older cars, damaging GATX’s residual values.

Speculative frenzy takes hold in some car type in the future: Used car lease rates have done best historically when demand for one particular car type takes up all available new car manufacturing capacity, forcing shippers to source equipment in the used market. This happened with ethanol tank cars in 2005, Bakken crude-by-rail cars in 2013-2015, and Canadian crude-by-rail cars briefly in late 2018-early 2019. It’s possible that some specific tank car application catches fire in the next few years which repeats this dynamic. However, we would note that tank cars in storage are 112k as of early December – up from 90k in February 2019 and near 2016 levels of 113k tank cars in storage – implying significant slack in the system that would need to be absorbed before this dynamic can take place. There appears to be around 35k-40k of annual tank car manufacturing capacity in North America, so the used tank car market becomes very tight when new tank orders are in the 8-10k units per quarter range, as was the case from May-December 2018. As the Canadian CBR opportunity is now being met with available tank cars, new orders dropped to 1.7k in the quarter ended September 2019.

Other notes:

Rolls Royce Partners Finance: We are aware that many bulls cite RRPF as a hidden gem asset inside GATX. For background, RRPF is a 50/50 Rolls Royce/GATX joint venture that leases spare engines manufactured by Rolls Royce. Airlines that use RR engines typically keep a number of spare engines on hand in their maintenance hubs in order to minimize aircraft downtime. For capital efficiency those spare engines are almost always leased by the airlines, typically from the Rolls captive finance company RRPF. The main risk to engine leasing is technological advancement, as used older-generation engine values typically decline precipitously when newer generation engines become available. Given that RR themselves govern the pace of next generation engine rollouts, we would expect them to manage the older generation residuals well. Hence, we assume a slightly growing earnings contribution from RRPF affiliate income, and value that income at the same multiple as the rest of GATX. However, such a valuation is debatable considering where other aircraft leasing businesses like AER/AL trade who serve the same customer base. Furthermore, RRPF is extremely levered with debt/EBITDA of 11x at year end 2018, and there are few specialty leasing companies with that type of leverage trading at a 15x P/E.

Catalyst

Earnings declining over time