TREX is over-earning due to a cyclical peak in home remodeling activity. Over 90% of Trex’s sales are derived from domestic residential remodeling. Remodeling activity is now back to the levels achieved at the peak of the last cycle (which in hindsight we know was a massive housing bubble) and TREX’s sales and profits are correspondingly elevated. Shares are expensive both on normalized and peak earnings.
Business Overview and History
There are essentially three options when building a new deck; Wood is the most common and cheapest. 100% plastic / PVC is the most expensive and least common. The third option (which is what TREX produces) is a composite of plastic and wood fiber. Plastic / composite decking can cost 2-4x wood decking, but offers the benefits of requiring lower maintenance and not needing to be stained. TREX’s products are sold through a nationwide network of regional distributors who in-turn sell to retail outlets. Trex manufactures its own polyethylene (50% of raw materials by volume) from recycled sources and recently expanded capacity in order to sell polyethylene to outside firms.
TREX was spun-out of Mobil Corporation in 1996 and went public in 1999. During the 2000’s synthetic decking took a significant share of the market from wood, although these share gains plateaued by the end of the decade. During the housing crisis in 2008-2009 a number of smaller players in the composite space went out of business. As shown in Figure 1 below, Trex was able to capture some of the business of these bankrupt companies causing the company to grow share from 30% in 2007 to 37% in 2011. During this same period the share held by smaller, independent manufacturers fell from 42% to 24%. In 2012 two of Trex’s largest competitors, Azek and TimberTech, merged. This merger experienced significant integration issues which allowed Trex to convert some distributors who previously carried Azek / TimberTech’s products, to Trex products.This further increased Trex’s market share from 37% in 2011 to 40% today.
Figure 1: Historical Market Share of Composite Decking
Remodeling Activity is at a Cyclical Peak
Home remodeling is highly cyclical in nature and tends to follow the strength/weakness in the overall economy. As shown in Figure 2, by any measure current levels of remodeling activity are back to the highs of the past cycle.
Figure 2: Remodeling Activity is at a Cyclical Peak
TREX is Over-Earning
Since over 90% of Trex’s sales are to the domestic remodeling market, it should not be surprising that sales are strongly correlated to overall remodeling spending. As proof of this relationship, a simple linear regression of Trex’s sales to total remodeling activity (as calculated by the Joint Center for Housing Studies) and the company’s market share yields the following equation with a highly statically significant relationship and an R-squared of 0.84 (a full regression output is shown in the appendix):
Using this equation we can calculate what Trex’s sales are likely to be in a more normalized remodeling environment. From 2004 – 2015 (a time period that includes a large housing bubble) the average annual remodeling spend was approximately $126B/yr. As shown in Figure 3, assuming the normalized level of future remodeling cycles is similar to historical levels and Trex’s market share is unchanged, we would expect normalized sales to be about $334mm.Even if you assume remodeling spending in future cycles increases to $135B/yr and Trex’s market share grows to 41%, normalized sales would be $369mm. Applying a justified multiple to the earnings of each of these scenarios yields a fair value of $12-$23 per share or approximately 45%-70% below current levels.
Figure 3: TREX Normalized Earnings and Valuation
Peak-Level Risk is Acceptable
Trailing 12 month remodeling spend is about equal to the peak of the last cycle. Even so, TREX still trades at 24x trailing P/E. While it is certainly possible remodeling spending could continue to grow to a level higher than the previous peak, it seems unlikely that the peak of this cycle will be significantly higher than the prior cycle when you consider that at the time of the prior cyclical peak the U.S. was in the middle of the largest housing bubble in the last 100 years.By my calculations (below) even if you assume some continued market share growth, for TREX to trade at an 11x P/E the remodeling market would have to grow by an additional 47%. As such, the downside risk here seems acceptable.
Figure 4: How Much Growth is Needed for TREX to be Cheap?
The Normalized Level of Home Remodeling has Increased – As is always the case with over-earning businesses, some would argue that ‘this time is different’ and that we are not currently at a cyclical peak for remodeling. There are generally two reasons given to support this view; the first argument is that current low interest rates, high-housing prices and availability of credit make remodeling a compelling investment. The problem with this argument is that while remodeling may indeed currently be a good investment due to high housing prices, low interest rates and easy credit, these factors are themselves cyclical in nature. Since there is no reason to believe that either house prices, interest rates or credit availability has undergone a secular change, this argument actually supports conclusion that remodeling investment is cyclically elevated.
The second “this time is different” argument contends that the U.S. housing stock continues to grow and age and as such requires more spending to remodel. While the U.S. housing stock is indeed older and larger than previous cycles, the amount of difference is small and not nearly enough to offset historic housing bubble experienced during the last remodeling cycle.
Furthermore, even if you do in fact assume that mid-cycle remodeling spending going forward will be higher than previous cycles, TREX is still significantly over-earning and significantly over-valued on mid-cycle earnings power. We can see this just by looking at the current earnings. With the remodeling market at all-time highs TREX still trades at 24x P/E which is much higher than any reasonable justified multiple. Thus, in order for TREX to be cheap on a mid-cycle basis, normalized remodeling activity has to be significantly higher than all-time highs which seems highly unlikely.
Increased Market Share– From 2007 to 2014 Trex has increased its market share of composite decking from 30% to 40%. While these market share gains are impressive, they are the result of two specific one-time events and there is no reason to think share gains will continue. First, the housing crisis allowed Trex to capture share from smaller players that went out of business. Then in 2012/2013 integration issues with the Azek / TimberTech merger allowed Trex to convert some distributors who previously carried Azek / TimberTech’s products, to Trex products. As such, it seems appropriate to view these market share gains as one-time in nature.
Furthermore, even if TREX gains significant additional market share, shares are still expensive. If we assume TREX’s share goes from 40% to 50%, mid-cycle sales and EPS would be approximately $435mm and $1.90 – implying a mid-cycle P/E of about 21.5x which is still higher than the high-level of justified P/E we calculated above.
Increased International Sales– Trex estimates that approximately 50% of the total market for composite decking is outside North America. Since 2010 Trex has been selling products outside North America, but so far international sales are a very small (and undisclosed) percentage of total sales. Given that in general international expansion tends to destroy shareholder value and thus far Trex’s international efforts have produced minimal sales, this risk appears acceptable.
Appendix: Regression Output
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.