Trinity Industries (TRN) is undervalued for two reasons:
1) Short-term earnings expectations from analysts and management are very low due to macroeconomic concerns justified by a significantly diminished backlog of orders and new order inflow. However, long-term earnings expectations are still strong based on industry fundamentals and independent forecasts for product demand. Even taking the most pessimistic scenario that I can concoct, TRN appears undervalued by over 50% at a present share price of around $10.
2) Earnings have been systematically understated due to the method of accounting for transfer of assets between subsidiaries. If we separate the railcar leasing division from TRN, the combined pro-forma earnings would have been between 10-30% higher over the past few years. This difference should continue as long as TRN builds its railcar leasing fleet, but if we froze the business today and simply let the assets generate revenue, shareholders would reap the benefits for up to 30 years.
In an efficient market, investors and analysts would be indifferent to the accounting method described briefly in reason (2). However, the market is either a) discounting future cash flows at an absurd rate or b) expecting absurdly low future cash flow, as described briefly in reason (1). These two factors seem to have a synergistic effect that has destroyed (in the eyes of the market) the investment value of what I believe to have been a wise strategic allocation of resources that smoothes capacity utilization and sets TRN up for strong future returns.
Background
Trinity Industries’ primary business is the manufacture of railcars. In 2008, 46% of total railcar production in the US came from TRN, and 49% of industry-wide orders placed in the 4th quarter went to TRN.
We can see from data provided by the Railway Supply Institute that railcar orders and deliveries since 1956 have been volatile with no clear growth trend. Average orders and deliveries over this time period have been 51,693 and 50,484 with standard deviations of 32,866 and 23,029 respectively. Because the industry operates with a backlog, we would expect the standard deviation of annual deliveries to be lower than that for orders. However, the high variance in deliveries indicates that competition for new orders is fierce, with capacity being added and removed in step with order demand.
We can compare annual sales data to the recent sales and forecasts from TRN. Forward estimates of industry unit sales are provided by TRN in their most recent 10K, courtesy of Global Insight, Inc. and Economic Planning Associates, Inc. I included future TRN unit sales as simply 1/4 of industry sales in the graph below, and consider this as one possible scenario discussed in the valuation section. We can see from this chart that recent sales have been above the 50 year average, and that sales are expected to return to slightly elevated levels by 2012 despite short-term economic weakness.
In recent years, sales to external clients have diminished as total industry sales have decreased since 2006 and TRN has been diverting newly manufactured railcars to a leasing fleet. External revenue has been $1,381, $1,540, and $1,516 million in 2008, 2007, and 2006 respectively. If we include intersegment revenue, TRN produced $2,563, $2,382, and $2,143 million worth of railcars in 2008, 2007, and 2006 respectively. We can see from this how TRN is smoothing production and capacity utilization in a volatile market. This strategy stores inventory and utilizes it at an attractive rate of return through lease agreements and eventual securitization.
The leasing division is known as the Trinity Industries Leasing Company (TILC). This company manages a fleet of railcars, occasionally securitizes pools of railcars, and owns a 25% equity position in TRIP Holdings, LLC, an affiliated leasing company with four equal partners (including TRN). TILC has increased its fleet from 20,300 railcars in 2004 to 47,850 in 2008. Revenue in 2008 was $536 million with $222 coming from sales of fleet railcars. EBIT was $158 with $35 coming from sales of fleet railcars. Of the sales of fleet railcars, $134 million of revenue and $21 of EBIT came from sales to TRIP, of which TRN retains a 25% interest. However, the gain on sales to TRIP does not account for the portion of EBIT that would have been recognized by TILC had TRIP not been partially owned. The result of this is a situation where TRN is penalized (relative to comparison companies such as ARII) for deferring revenue with lease agreements or sales to TRIP regardless of the favorability of the lease terms. Debt incurred by the lease group has traditionally been non-recourse, meaning that TRN is generally not responsible for the interest payment shortfall of securitized lease subsidiaries. The securitization process allows TRN to effectively sell the railcars to a group of asset-backed security investors who are otherwise not part of the rail industry, and potentially continue to earn a spread on the interest expense and leasing rate without having to tie up equity.
TRN also manages a few other businesses that are categorized into three reporting segments. An inland barge segment generated $625 million of revenue and $119 million of EBIT. This segment has grown at a rate of over 25% for the past two years, and the backlog remains at almost a full year’s worth of production, but management expects a decline of $6-16 million in 2009.
An “energy equipment group” combines a pressurized tank and tank head business with a structural wind tower business, and was responsible for $633 million of revenue and $100.3 of EBIT in 2008. Many of the tank heads produced by this business are used in liquid or fuel transport railcars, but it also produces fuel tanks ranging in size from 9 to 1,800,000 gallons. The structural wind tower business has exploded in recent years, and managed to grow the backlog from $700 million at the end of 2007 to $1.4 billion at the end of 2008, enough to sustain current level of business for 3 years without any new order flow. However, management has indicated an expected decline in revenue from this business of about $50 million in 2009 as a result of project delays.
Lastly, TRN operates a construction products business that sells cement and asphalt in Texas and highway construction safety products such as crash guards and safety rails nationally. This segment produced $741 million of revenue and $58.2 million of EBIT in 2008, but has not exhibited significant growth. Management made no specific estimates for this business segment in 2009, but expect downward pressure for at least the first 6 months.
Industry Economics
The macroeconomics of the rail industry are mixed at best. As was shown above, the shipments and new orders in recent years were above the historical average by about one standard deviation. If I had taken the average since 1980, the average would have been much lower, but we would see evidence of a growth trend. However we look at it, it seems like recent performance and short-term future expectations don’t deviate by more than one standard deviation. I have read reports with industry experts giving guidance as low as 12,000 units produced in 2010, to which I would refer to the extreme volatility in orders as proof that forecasts are difficult to make even a few months out.
Breaking orders down into finer categories, we see that coal-transport has been about flat, transport related to the ethanol industry has been decimated, and intermodal transport (freight that uses at least two modes of transportation, usually rail+truck) has poor prospects as a result of a generally weak economy and low consumer spending.
Another factor to be aware of is railcar age. On the one hand, these railcars last 30+ years, so the replacement market churns slowly. This is part of the reason why we see such volatility in production and order numbers. It’s a capital expense that can be put off until buyers have better visibility. On the other hand, TRN reports in the most recent K that “the average age of the North American freight car is approximately 19.3 years, with over 39% older than 25 years”. I’ve seen similar numbers confirmed elsewhere.
Of all the potential macroeconomic factors that could potentially impact the long-term future of rail, I think the cost efficiency of shipping freight is probably the most important. In an economic feasibility study done for the Transportation Agency for Montery County in September, ’08, Dr. Johnathan Mun concluded that intermodal rail shipping “indicates significant savings to the shipper-growers, providing a higher level of profit margin and reduces the variability in revenue over time, as well as providing a legitimate alternative option in terms of transporting produce to the market”.
Of course, economic feasibility doesn’t immediately translate into strong demand for manufacturers. There are also short-term concerns that decreased consumer spending can lead to less global trade and more goods being made near where they are used. However, it stands to reason that long-term trends will favor efficiency, which rail can certainly be a part of.
Debt
Total debt outstanding at the end of 2008 was $1,906 million, up from $1,374 in 2007, with an interest coverage ratio of 5.5x. However, $1,190 million of this debt was non-recourse to TRN compared to $644 in 2007. This arrangement actually transfers risk from TRN to the lenders, as discussed briefly by Seth Klarman in his 2007 address at MIT. If we consider only the recourse portion of the debt and the non-leasing EBIT, the 2008 interest coverage ratio rises to around 13.3x.
Of the recourse debt outstanding, $61.4 million is due in 2009, and the rest is due in more than 5 years. The non-recourse debt has been separated into three separate issuances. TRL-V and VI are wholly owned subsidiaries that own and lease railcars under 30-year financing agreements totaling $877.6 million. The remaining $313 million non-recourse debt is classified as the “warehouse facility” which matures in August 2009. Management indicated in the most recent conference call that they intend to renew this agreement, but if they fail to do so, the outstanding balance must be repaid in three equal installments between February of 2010 and 2011. However, even in the worst-case scenario, this debt could be repaid by surrendering the railcars collateralized in the agreement.
Earnings Adjustments
TRN accounts for earnings in such a way that growing the railcar fleet reduces the revenue and earnings relative to what could be booked if TILC was a separate business. The effects of this amount to what could have been an additional 10-30% of NI over the past few years.
As it stands now, the railcar manufacturing segment books revenue and EBIT from sales to external customers. TILC books revenue and EBIT from leasing agreements as well as the sale of fleet cars. Intersegment sales that grow the lease fleet are partially accounted for as lease subsidiary CapEx and are approximately equal to intersegment revenue minus “subsidiary EBIT eliminations”, meaning they are booking the cost of producing railcars as CapEx. The EBIT that would have been earned on sales to outside customers is left unaccounted for until TILC sells railcars, at which point a gain is recognized on the sale because the sale price exceeds the book (CapEx) value.
If TILC was a separate business, the cost of growing the lease fleet would be fully accounted for as CapEx and TRN would book sales to TILC at fair market prices. The effect on NI each year is equivalent to the after-tax lease subsidiary EBIT elimination minus the gain on the sale of railcars from TILC. The same effect hits FCF as long as maintenance CapEx is separated from growth CapEx. If total CapEx is grouped together (which it shouldn’t be because TRN can cut the growth instantly and continue operating without any loss of revenue for almost 30 years) then the effect disappears as the benefit to CFO is erased by an increased lease group CapEx.
The way TRN accounts for revenue, EBIT, and CapEx is appropriate. It’s just important to note that they are trading earnings in the current period for a stream of revenue that is expected to play out over as many as 30 years. In a normal market, analysts should be indifferent between high earnings today with low growth prospects or low earnings today with higher growth prospects. But because of the current market turmoil and a poor short-term earnings forecast, it seems as though the market is completely discounting the long-term effect of what was probably a very smart decision by management.
Another way to make this adjustment would be to include the issuance of debt from securitized lease fleets as cash flow from operations, which makes sense because the risk of ownership has been transferred to an unrelated third party. But this method creates lumpiness and ignores railcars that have not yet been securitized or “paid for” by the warehouse facility.
Separating the leasing segment also helped me reconcile an unrelated concern. I noticed when I first started looking at this company that the depreciation rate has steadily declined from 9.6% in 2002 to 4.7% in 2008. This made me concerned that depreciation was being under-accounted for and earnings were therefore overstated. However, as the railcar fleet grows as a percentage of total PP&E, we should expect the average depreciation rate to decline. If we subtract the depreciation that could be attributable to the railcar fleet and compare this to the non-leasing PP&E, we get a steady depreciation rate of around 20%.
Why I Love the Leasing Business
TRN builds a railcar and it sits as inventory on the balance sheet until management chooses from one of the following options:
1) To sell - the balance sheet trades inventory for cash.
2) To lease – the railcar moves from short-term inventory to long-term PP&E, and TRN expects a stream of revenue for the next 30 years.
3) To lease & finance - the railcar moves from short-term inventory to long-term PP&E, TRN gains the same amount of cash as in a sale, and the stream of cash from leasing should just about equal depreciation and interest payments.
4) To lease & finance with non-recourse debt - the railcar moves from inventory to PP&E, TRN gains the same amount of cash as in a sale, and the stream of cash from leasing is used to cover depreciation and interest payments. If the railcar becomes more valuable in the future, TRN benefits from a higher cash flow as with other lease options. However, if the railcar becomes impaired, TRN can choose to forfeit the railcar in lieu of interest or principle repayment.
Every railcar in the lease fleet that has been financed with non-recourse debt is like a call option on the railcar industry. There is a downside limit to carrying value losses that comes from the financing principle that doesn’t have to be repaid. Upside is unlimited. These options were issued “out of the money” because the asset value at the date of securitization was greater than the debt principle. But given the volatility of the railcar market, my guess is that these options are extremely valuable. Building this lease fleet was a very smart move.
Valuation
To value this company, I’m going to ignore the earnings adjustments made above and model a pessimistic forecast for this business based upon analyst estimates from Bloomberg and management forecasts from the most recent conference call. It will probably be easier to follow this with the excel model open.
I estimate revenue for each segment individually. From the rail group, I assume TRN continues to produce half of all industry orders, but that half of those orders are used in the lease fleet, leaving a quarter of industry sales to generate revenue. I estimate average unit sales price for a railcar in two ways. First I compare the backlog units to the provided backlog value from 2004-2008 to get an AUSP of $79k. I also compare the external revenue to external unit sales for the same time period to get an AUSP of $83k. To estimate sales from the rail group, I take a quarter of forecasted industry demand at $80k per unit to yield revenue in 2009 of $566 million. This number changes each year with forecasted industry demand. Revenue from the leasing segment is calculated as the 2008 fleet of 47,800 railcars leased at $7,582/year, the rate I estimated as average for 2008. The rate I estimated as average from 2004-2008 is $7,646, which seems reasonable as a railcar RoA before maintenance expenses. This yields a constant revenue stream of $363 million, which I assume as no growth because I model CapEx from the leasing segment as zero to demonstrate the cash flow characteristics of the business. No guidance was given on the construction group, so I model that as a flat $741 million going forward. Management offered a pessimistic forecast of $160 million ($170 optimistic) from the barge group carried forward with no growth. And the energy segment is modeled as experiencing a decline of $51 inline with management expectations for the structural wind tower business despite a backlog that could support 2008 production levels for at least 3 years.
This yields revenue in of $2,891 in ’09 and $2,799 in ’10 compared to analyst forecast from Bloomberg of $2,937 and $2,789.
I only model two main cost lines, COGS and SG&A. SG&A varies with revenue 0%, meaning I assume all costs are fixed. COGS are modeled as having a variable and fixed component. If I set the variable portion of COGS to 43%, my EBIT line in 2009 matches analyst forecasts, so this is where I have it set. Increasing the variable portion increases EBIT as costs can be taken out more quickly in a bad economy. Steel is variable cost, labor is less so, and office leases not at all, but in any case this is a rate implied by the market. I had very little to do with determining what it should be.
From $169 million analyst expected EBIT, they yield $81 net income. I’m not sure how this is possible with interest expense of $100 million in 2008. I expect debt outstanding to decrease from $1,905 by $61 to $1,844. However, this isn’t significant enough to allow $81 million of NI unless TRN pays no taxes. So my NI turns out to be lower than analyst expectations, but this is partially made up for in EPS, as I expect some serious share repurchases in 2009. Management announced a plan to repurchase $200 million in stock, equivalent to more than 1/5 of the shares outstanding at recent prices, of which $140 is left. If we divide analyst expected NI of $81 by the expected EPS of $0.99, we see implied shares outstanding of 82 million. That’s higher than the current diluted share count. So I have shares outstanding of 69 million after $140 million of repurchases at ~$13.50/share, significantly higher than where the shares have been trading. If I don’t limit share repurchases to $140 million, my model estimates repurchases of $217 because that’s how much cash is available (over what the company generally keeps on the books) and repurchases is by far the most useful allocation of cash.
The graph above shows how analyst expectations, back-of-the-envelope estimates, and my model compare to each other. The model guesses what revenue will be as a function of PP&E. We can see that my model’s revenue predictions are conservative, so I’m comfortable using these numbers going forward. I also included the EBIT margin predicted my model. This is a function of historic COGS and SG&A levels as a function of revenue, as well as industry growth (which is 0% for all but the first period) and the variable COGS which only effects the first period. Revenue is low in ’10 because CapEx wasn’t great enough in ’09 to cover the cost of depreciation, reducing PP&E. This is probably not inaccurate as a representation of the effects of cost-cutting measures that would have a negative impact on future earnings unless there is industry growth.
In the final period, I compute the share price as a 9-10x multiple of 2013 earnings. This is an average multiple since 2005 when earnings turned positive, and represents a discount rate minus growth rate of around 11%. The share price for each period before that is the dividend per share plus the value of next years expected share price discounted at 15%.
The model also doesn’t allow for new issuances of debt or growth in the lease fleet, and $11 million is spent on acquisitions each year that don’t benefit the company in any material way. Despite this, the model’s EPS figure is higher than analyst expectations in ’10 and ’11 as a result of heavy share repurchases at a deeply discounted price in ’09. I set the maximum share repurchase to $140 million, inline with what management has been cleared to repurchase, which would have been exceeded based on excess cash reserves if there was no limit. The result is an additional repurchase the following year of $65 million shares at a less favorable price. Using this pessimistic earnings forecast, my model estimates a current share value of around $16 today and $30 in 5 years after dividends.
Comparison
TRN is also undervalued relative to its peers. For starters, it’s the only company that invested heavily in building its own leasing fleet. This hurt current earnings for TRN relative to its peer group in current quarters, but built an asset base that will at worst generate some revenue and at best appreciate in value and be available for sale at a premium.
FreightCar America (RAIL) is TRN’s largest competitor by market cap and the smallest by enterprise value with $10 per share in cash. It’s also the only competitor that has begun to retain railcars in a lease fleet. However, even if we make adjustments to earnings similar to those described above for TRN, the EV/EBIT multiple for RAIL is still 6.5x compared to TRN’s 4.4x. On a forward multiple basis, RAIL appears grossly overvalued unless we take out the cash, at which point we get forward multiples of about 15x and 11x in ’09 and ’10, but this compares to multiples of 12x and 10x for TRN. Qualitatively, RAIL is heavily dependent on demand for coal transport. TRN has a far more diversified business in terms of both railcars and some other important business segments not related to rail at all. Just for fun, I’m looking at the forward curve for coal (API2 on Bloomberg), and the yield on keeping coal in the ground is over 10% through 2013 (rising from $66-98). So obviously the demand for coal is expected to increase, but a) this is a European index and b) I’m not sure what kind of effect it would have on RAIL stock even if it was US futures.
American Railcar Industry (ARII) is another railcar manufacturer owned 54.4% by Carl Icahn who also owns a non-consolidated railcar leasing business (ARL), so ARII doesn’t benefit at all from the earnings adjustments made above. ARL is responsible for $133 million out of $373 million total of backlog. Icahn also owns ACF, an unconsolidated business that trades everything from raw materials to entire railcars with ARII. These arrangements make me uncomfortable. I wouldn’t consider investing in this company unless I had a lot of time to get comfortable with management and fully understood the Icahn related arrangements, and I wouldn’t contact management unless ARII was obviously the undervalued company in this industry, which it’s not based on forward earning multiples. I’m not going to dig my teeth too deep into this porcupine.
Greenbrier Companies (GBX) superficially appears to be the most undervalued competitor with a LTM FCF multiple of 1x and forward earnings multiple of 6x in ’10, compared to 10x for TRN. A recent Q shows a large disparity between earnings and FCF, so figuring out why is probably a good place to start. It looks like in the first half of this year (two most recent quarters) margins have simply been squeezed. There were no one-time items or write-downs to take out of earnings, no unusual changes in interest expenses, but earnings came in at about $5 million, compared to a historic range of around $20. On the cash flow side, it does look like GBX had an unusual flow of cash from a reduction in working capital that had been built up 5-6 quarters ago. So I don’t think either are necessarily an accurate representation of what GBX normally does or will do in the future.
Another concern is that General Electric is currently trying to back out of an agreement to purchase 11,900 railcars, compared to a total backlog of 15,100 units at the end of the most recent quarter. GE might be stuck having to take delivery of that order which is good for GBX, but it still must be an uncomfortable strain on a relationship with a major customer. Also, compared to TRN, GBX has less favorable debt which mostly obligates the parent company to guarantee subsidiaries. EBIT/interest is 1.25 LTM, and it dropped below 1 for the first half of ’09.
Conclusion
It seems to me that if the future looks anything like the past, TRN is grossly undervalued. And I don’t think that the railcar industry or the long-term prospects for the global economy have or will be changing enough to suggest that the future for rail will be worse than the past. TRN is undervalued relative to it’s peers, it has built a valuable asset in it’s lease fleet, long-term macroeconomics are favorable, and my pessimistic no-growth scenario has TRN undervalued by 60% if investors demand a 15% return for the next 5 years and a 10% return for eternity.
Disclosure: None