Cloud Infrastructure

Wireless Tower Valuation Update



By Catalyst Research Team

1) Summary of Findings

  • We investigate the determinants of wireless tower company valuations, employing the EV / LQA EBTIDA multiple of an enterprise value-weighted public tower company index as the dependent variable.  We explore a variety of independent variables including interest rates, credit spread, curve structure, and broad market valuations, along with other macroeconomic factors.

Tower Valuations Have Risen Since the 2008 Trough

Monthly (n=157) Valuation (EV / LQA EBITDA, Light Green) of Enterprise Value-weighted Custom Wireless Tower Company Benchmark, Average (Dark Green), and +/- 1 SD Confidence Interval Bars (Red)

Screen-Shot-2016-10-12-at-10.59.02-AM

Source: Capital IQ, Catalyst analysis.

  • First, we find that tower valuations are linked to interest rates, but that this relationship is unstable over time.  The directionality of this relationship has changed: a positive relationship existed between the two variables from 2002 until an inflection point coincident with the 2008 financial crisis, followed by a negative relationship thereafter.  However, this transition is also characterized by a number of exogenous events including Quantitative Easing, the maturation of the tower sector as a whole, and the REIT conversion of AMT and CCI in 2012 and 2014 respectively.  We find the strongest long-run relationship with tower valuations at the short and long ends of the yield curve.

U.S. Treasury Interest Rate Correlation to Tower Company Benchmark Valuation

2 Year

5 Year

10 Year

20 Year

2010 – 2015 R

-0.19

-0.19

-0.36

-0.45

2010 – 2015 R2

0.04

0.04

0.13

0.20

2005 – 2010 R

0.77

0.76

0.76

0.71

2005 – 2010 R2

0.59

0.58

0.57

0.51

13 Year R

0.49

0.37

0.25

0.09

13 Year R2

0.24

0.14

0.06

0.01

Source: Capital IQ, Catalyst analysis.

  • Second, we find that tower valuations are linked to credit spread, particularly high yield spread and that of public utilities.

Credit Spread Correlation to Tower Company Benchmark Valuation

Low Investment Grade

High Yield Proxies

BAA Corporate Spread

BAA Public Utility Spread

BB High Yield Spread (Goldman)

HYG – AGG

Spread (1)

13 Year / Max R

-0.40

-0.50

-0.58

-0.84

13 Year / Max R2

0.16

0.25

Not Reported

0.70

Source: Capital IQ, Goldman Sachs, Catalyst analysis.

(1) Limited timeframe (n = 31) may bias results.

  • However, we contend that spread is a proxy for market appetite for corporate risk, rather than a determinative variable by its own merit.  Spread has a strong and persistent inverse relationship with interest rates, which suggests that while the effects of rates and spread on valuations occur in tandem, their relationship is unclear.  We also find more robust alternative measures of market expectations than credit spread.

Macroeconomic Factor Correlation to Tower Company Benchmark Valuation

Macroeconomic

Market Sentiment

S&P 500

Inflation

Real GDP

Unemployment

New Orders

Consumer Confidence

Total Return

CAPE Ratio

CPI

(YoY %)

TIPS Spread

1 Year R

0.49

-0.29

-0.46

0.26

0.48

0.39

-0.62

-0.58

1 Year R2

0.24

0.09

0.21

0.07

0.23

0.15

0.38

0.33

13 Year R

0.33

-0.49

0.45

0.45

0.37

0.48

0.39

0.49

13 Year R2

0.11

0.24

0.20

0.20

0.13

0.23

0.15

0.24

Source: Capital IQ, Robert Shiller, Catalyst analysis.  TIPS spread based on 20 Year node.

  • Third, we introduce curve structure as a new variable, and find a strong relationship between slope and tower valuations.  This sensitivity to the relationship between short and long rates suggests that tower companies should benefit in steeper curve environments.  From the perspective of tower companies, short rates determine the cost of debt financing and relate to the potential for accretive acquisitions, while long rates point to a substitution effect between towers and treasuries as long-duration income-producing assets.  We find these metrics to be robust and consistent, and focus on the 2s20s slope in our analysis.

Curve Structure Correlation to Tower Company Benchmark Valuation

2s10s

2s20s

2s30s

5s10s

5s20s

5s30s

1 Year R

-0.48

-0.45

-0.43

-0.35

-0.33

-0.31

1 Year R2

0.23

0.20

0.18

0.12

0.11

0.09

13 Year / Max R

-0.60

-0.63

-0.63

-0.56

-0.61

-0.60

13 Year / Max R2

0.36

0.40

0.40

0.31

0.38

0.36

Source: Capital IQ, Catalyst analysis.

  • Finally, we conclude that while rates, spread, curve structure, and other macroeconomic proxies are all related to interest rates, the persistence and directionality of these relationships, especially to benchmark rates, are highly variable.  We consider a blanket statement that interest rates relate directly to tower valuations to be inaccurate, and acknowledge that the recent 13 year period of available data we analyze may also reflect changes owing to exogenous trends including maturing industry fundamentals, increasing internationalization, and changing carrier capital expenditures, among others.
  • The relative momentum of trends in rates, as noted here, and wireless tower market dynamics, as noted in Catalyst’s Industry Update concurrent with this memorandum, are likely to persist in the short run.
  • Considering the factors analyzed on a combined basis, we hold a bullish view on wireless tower company valuations.  However, there is a high degree of uncertainty as to the timing and nature of rate movement following the end of Quantitative Easing, and the various scenarios outlined below should be considered with discretion and subject to the limitations previously discussed.
    • Interest Rates: In a parallel shift of interest rates upward, the long run data indicates either no change or modest appreciation in tower multiples.  Observations drawn from the recent period from 2010 – 2015 would suggest multiple contraction, while the period from 2005 – 2010 would support multiple expansion.
    • Credit Spread: If corporate and / or high yield spread were to increase, the long run data suggests a decline in tower multiples.  However, we believe that spread may not be an independent variable in its own right, on the basis of its strong and persistent relationship to both interest rates and general macroeconomic variables.
    • Macroeconomic Factors: If macroeconomic fundamentals were to deteriorate, we would expect a normal contraction in market multiples, including but not limited to wireless towers.  We find evidence that wireless towers exhibit higher-than-average procyclicality, or that their valuations are more sensitive to cyclical factors than those of the market at large.
    • Curve Slope: The shape of a changing yield curve is difficult to predict, as each node will rarely shift an equidistant amount.  In particular, we would expect multiples to decline in a low or negative slope environment, such as a scenario in which short rates rise more than long rates, or a “bear flattener.”
    • Other Factors: The REIT conversion of leading tower industry participants has led to the inclusion of AMT and CCI among REIT indices and investment mandates.   If a broad REIT selloff were to occur (due to valuations exceeding fundamentals), technically-driven selling of REIT instruments (such as mutual funds and ETFs) could lead to deterioration in tower industry multiples.

2) Background to Analysis

What aspects of wireless towers are most important to our analysis?

  • Wireless towers are unique assets that bridge the gap between communication infrastructure and real estate – just as a data center is a parking garage for servers, so too is a tower an apartment building for wireless transmitters.
  • Tower industry growth will continue to be supported by strong secular trends toward increased data usage, wider international adoption of wireless technology, limited new tower construction, built-in lease escalators which generally exceed the rate of inflation, the extension of operating leverage through scale, and the potential for accretive industry consolidation.  However, wireless towers also entail technological risk and are much less easily repurposed for alternative uses than traditional real estate.  There are three public companies in the space: American Tower (AMT), Crown Castle International (CCI), and SBA Communications (SBAC).
  • Investors generally seek growth, income, preservation of principal, or a combination of the above.  Growth, or capital appreciation, is typically achieved through exposure to equities.  While bonds are the traditional fixed income investment, MLPs, REITs, and unconventional fixed income instruments also provide investors with high income yield.  Preservation of principal is generally achieved through exposure to limited duration fixed income, inflation protection, and alternative strategies.  Of these three allocation goals, wireless tower companies offer investors a unique combination of growth and income, and may choose to exhibit more “growth” or “income” characteristics by either increasing their leverage or dividend payout ratios, respectively.

What’s so interesting about interest rates?  What do we expect in the future?

  • Following a multi-decade bull market in bonds characterized by falling interest rates, U.S. yields are near historic lows after beginning to rise in April 2013 (the market’s “taper tantrum”) and then retrenching throughout 2014.  The Fed has indicated the earliest date by which the Fed Funds rate could be raised is the end of Q2 2015.  While the future is uncertain, it is believed rates will rise following diminished open market operations and a rising fed funds rate.

Treasury yields remain low relative to norms, within 25 bps of absolute minimum

10 Year U.S. Treasury Yield (Light Green), Average, (Dark Green) and +/- 1 SD (Red) 68% Confidence Interval (Trailing 5 Year)

Screen-Shot-2016-10-12-at-11.03.03-AM

Source: Capital IQ.

  • While we (n=11) have predicted the 10 Year at approximately 2.5% as of YE 2015, Goldman Chief Economist Jan Hatzius expects it to reach 3.0% over the same timeframe.  Either would represent a substantial increase over current yield levels, around 1.8% as of 1/30/15 but rising through February.  To meet these forecasts, the yield curve would need to shift upward between 70 – 120 bps, essentially reverting to its level as of a year ago.  Curve shifts always affect each node (maturity profile) differently.

Annual Change in Curve (Last Five Years)

chart2

Source: Capital IQ.

Why might this matter to tower valuations?

  • In fixed income, there is an inverse relationship between price and yield: when yield goes up, the price of existing bonds goes down because new capital can be invested at a higher rate, reducing demand for lower-yielding bonds of comparable maturity issued at lower rates.  There are significant linkages between the fixed income and real estate markets, due in large part to the industry’s reliance on debt financing (the mortgage market where securitized).  Since real estate valuations are so closely linked to the cost of financing (indirectly determined by the Federal Reserve), rising interest rates have historically negatively impacted real estate valuations.

REIT Valuation and Interest Rate Relationship

Top: REIT Index Valuation (Red, Left Axis) and 20 Year Treasury Yield (Light Green, Right Axis)

Bottom: Rolling 1 Year (Dark Green) and Smoothed 3 Year Correlation (Orange)

chart8

Source: Capital IQ, Catalyst analysis.

  • This is likely due to a combination of increased borrowing cost (which decreases the aggregate level of new investment) and depressed real estate asset valuations (as demand for properties decreases after financing has become prohibitively expensive).  This theory suggests an inverse relationship between interest rates and tower valuations – which, if correct, would imply rising rates could negatively impact tower valuations.
  • We also might expect wireless tower valuations to be affected by other interest rate-related factors as well, such as credit spread (reflecting market appetite for corporate risk) and / or curve slope (reflecting relative relationships between different maturity nodes of the yield curve).  We expect the highest levels of sensitivity to interest rate changes at the short (2 to 5 Year) and long (20 to 30 Year) segments of the curve.
  • These factors lead us to contend that wireless tower assets have knowable implied maturity and duration, or relative sensitivity to rates, which relates to both their borrowing cost (or short rates) and the maturity of their tower lease portfolios (or long rates).  However, QE is an experiment – the effects of its unwinding are unclear, and over the long run, the behavior of conventional REITs during rising rate environments has been highly variable (as rates may rise for a variety of reasons).  Before 2008, the United States had never existed in a persistent low interest rate environment of the same magnitude.

How should we properly sensitize our investment case with these conclusions?

  • Arrive at probability-weighted forecast of rates, spread, slope and other key variables.  Combine forecast with sensitivity of valuations to rates, spread, and / or other key variables to arrive at probability-weighted expected change in valuation due to rates.  Impact may be a premium / discount factor to current valuation or more likely a change in projected exit multiple.
  • We will also consider which sector groups we feel are or may be most like towers in the future – including data centers, hosting, telecom, outdoor advertising, and traditional REITs.

Historical Growth of Sector Benchmarks

Total Enterprise Value of Custom Indices in $USD Billions, Trailing 13 Year, Logarithmic Scale, Between $1 Billion and $1 Trillion

Screen-Shot-2016-10-12-at-11.05.49-AM

Source: Capital IQ, Catalyst analysis.

3) Analysis of Third Party Research

Cowen & Co. (October 2014)

  • Find that the relationship of tower stock performance to interest rates is weak (R2: 0.44 to AMT, 0.28 to CCI, and 0.16 to SBAC), but growing, and that this correlation is expected to grow further.  Also argues that REITs are more sensitive to interest rates than non-REITs.
  • Methodology:  Compares monthly performance of individual tower stocks to the basis point change in the 10 Year Treasury over the trailing five year period (n=60).  As we do not have access to the full Cowen research, our conclusions are only based on the available information.
  • Catalyst Assessment
    • Cowen analyzes tower stocks in isolation, whereas we would construct a benchmark to proxy for the wireless tower industry as a whole.  We would consider correlation in addition to R2 as a statistical measure.  For the dependent variable, we would prefer to use a valuation metric, rather than monthly stock performance.
    • We consider the monthly bps change in the 10 Year Treasury to be a subpar measure of interest rates.  While change is a better indicator than absolute level for rates, taking monthly change in bps may introduce a level of noise resulting from normal volatility, which could distract from meaningful conclusions.
    • We do not agree that the relationship of rates to tower stocks is weak.  We agree that the correlation of rates to tower valuations has been increasing in magnitude and may continue to do so in the near term.

Corvex Management (October 2014)

  • Find a minimal (R2=0.15) but positive relationship between tower multiples (EV / NTM EBITDA) and interest rates, believes that the “fundamental interest rate sensitivity of the tower sector is often overstated,” and that market overreaction to rate-rise-related news could present investors with a buying opportunity further on.  Argues that towers have traded at higher multiples than today’s in higher interest rate environments, such as 20x – 24x NTM EBITDA between mid-2005 and 2007.
  • Corvex is an activist investor in CCI that wants the company to increase either dividend payout (more income) or leverage (more growth).  Corvex argued that the company is doing neither, and would prefer that leverage remain constant while dividend payout is increased, boosting appeal to income-seeking investors.  Following the whitepaper’s release, CCI rose from $80.8 / share to $84.8, pulled back through December, and finally closed at $86.5 on January 30th.
  • Methodology:  Compares weekly average tower company multiples based on forward earnings to the yield of the 10 year U.S. Treasury over ten years (2004 – 2014).

Corvex argues that rates and tower valuations are unrelated over the long run…

Screen-Shot-2016-10-12-at-11.07.14-AM

Source: Corvex Management.

But timeframe selection may unintentionally bias results – when looking at correlation on a rolling basis, they do not appear unrelated but instead related in opposing ways across two time horizons

Rolling 1 Year Correlation of 10 Year Treasury Yield to Weighted Average Tower Company Benchmark Valuation (EV / LQA EBITDA), Trailing 13 Year with 5 Year Averages (Red: 1/30/10 – 1/30/15, Light Green: 1/30/05 – 1/30/10, Dark Green: 1/1/02-1/30/05)

Screen-Shot-2016-10-12-at-11.08.06-AM

Source: Capital IQ.

  • Catalyst Assessment
    • Time period selection may bias this study toward neutrality, as the sign of the relationship between interest rates and tower valuations changed during the study period.  The past ten years includes five with positive correlation between the two variables and five with negative correlation.
    • We agree that “the fundamental interest rate sensitivity of the tower sector is often overstated,” and would add that it is not widely appreciated that towers and rates have experienced both positive and negative relationships of equal magnitude over the past decade.
    • We also agree that the market will likely overreact to any interest rate-related news and that this technically-driven selling may reduce the valuation of towers in the short run, along with REITs and other income assets.
    • We would argue that the observation on high tower multiples between 2005 and 2007 is somewhat misguided, as this period preceded the most substantial multiple contraction in the wireless tower market to date (concomitant with widespread multiple contraction in the broader market as well).  However, we agree that there is room for further appreciation at current valuation levels. 

Goldman Sachs (September 2014)

  • Find no (0.02) correlation (R) between tower valuations and interest rates, as represented by the inverse of the 10 year treasury yield.  However, find a strong (-0.58) correlation between tower valuations and high yield credit spreads.  They conclude that investors don’t buy tower stocks for yield, but for growth (recurring FCF or AFFO/share).
  • Methodology:  Compares valuation (EV / LQA EBITDA) of blended tower benchmark (AMT, CCI, and SBAC) to inverse 10 Year Treasury and High Yield Credit Spreads over trailing 13 years.
  • Catalyst Assessment
    • We agree that EV / LQA EBITDA is a better measure of value than monthly stock price and that credit spreads are an important element to incorporate in this analysis.  However, we question the significance of spreads in isolation, and believe they are being used to proxy for cyclicality, as spreads widen in risk-on and tighten in risk-off environments.
    • On the basis of low R2 compared to other variables (including other key macroeconomic variables) in the regression, it is likely that a direct line of causality does not run from credit spreads to tower valuations, and that valuations are instead normally correlated to market risk, or beta, along with other factors.  We agree that investors are likely attracted to tower stocks on the basis of growth rather than income, but we do not believe that this relates directly to rates or spreads.
    • The choice of the 10 Year U.S. Treasury Inverse to represent interest rates is poor.  It is not customary to take an inverse of an interest rate, and we reject their conclusion that interest rates and tower valuations are unconnected on the basis of this data.

Goldman finds little correlation between interest rates and tower valuations

Forward 12-Month EV / EBITDA Multiples for Towers vs. the inverse of the 10 Year Treasury Yield

Screen-Shot-2016-10-12-at-11.09.10-AM

Source: Goldman Sachs.

However, they find a strong correlation between tower valuations and credit spreads

Forward 12-Month EV / EBITDA Multiples for Towers vs. High Yield Credit Spreads

Screen-Shot-2016-10-12-at-11.10.42-AM

Source: Goldman Sachs.

4) Catalyst Primary Research

  • The Cowen, Corvex, and Goldman studies highlight the range of conclusions that can be drawn from similar data relating to the same underlying variables.  Below, we show a variety of interest rate metrics based on the 10 Year U.S. Treasury.  Not all are created equal: the inverse and monthly bps change stand out as particularly poor metrics over the long run.

Independent Variable Selection May Bias Third Party Results

Various Interest Rate Metrics Based on the 10 Year vs. Tower Valuations

Correlation

R2

1 Year

13 Year

1 Year

13 Year

10 Year U.S. Treasury Yield (%)

-0.63

0.25

0.40

0.06

Inverse of 10 Year U.S. Treasury Yield (normalized %)

0.60

-0.15

0.36

0.02

Monthly Change in 10 Year Yield (bps Δ)

-0.56

0.11

0.32

0.02

Annual Change in 10 Year Yield (bps Δ)

-0.51

0.31

0.26

0.10

10 Year Swaps (%)

-0.60

0.27

0.36

0.07

Annual Change in 10 Year Swaps (bps Δ)

-0.46

0.41

0.21

0.17

Source: Capital IQ, Catalyst analysis.

  • The substantial difference in correlations over the long term compared to short run suggests an unstable relationship and low statistical significance.  As earlier observed, time period bias may also significantly influence results, which leads us to consider rolling correlations across a variety of economic environments.
  • We specify our dependent variable as the enterprise value-weighted average valuation of public wireless tower company stocks, measured as the ratio of enterprise value to LQA EBITDA.  We employ enterprise value-weighting to construct our index on the basis that this better represents the valuation and performance of the underlying assets than market capitalization-weighting.
  • A recent Barron’s study supports this methodology and argues that given the real estate market’s heavy reliance on debt financing, enterprise value best represents the full economic weight of a company’s aggregate real estate capital stack (including debt, minority interest, and preferred).  They also found that enterprise value-weighted REIT indices were better able to bounce back from the 2008 financial crisis than those weighted by market capitalization, more accurately capturing the real estate market’s rebound.  We employ the same methodology to construct comparable indices for valuation in sectors which we believe have been or will be similar or related to wireless towers.

Correlation Matrix of Sector Valuation Indices

S&P CAPE

Towers

REITs

Telecom

Hosting (1)

Billboards (1)

Data Centers (2)

S&P CAPE              
Towers

0.48

           
REITs

0.02

0.47

         
Telecom

0.73

0.42

0.06

       
Hosting (1)

0.41

0.36

0.20

0.05

     
Billboards (1)

0.47

0.43

0.21

0.38

0.23

   
Data Centers (2)

0.50

0.28

-0.28

0.41

0.14

0.45

 

Source: Capital IQ, Robert Shiller.  (1) Index inception 11/30/05.  (2) Index inception 1/31/2004.

Catalyst Custom Benchmark Constituents

As of 1/31/15

Company Ticker Benchmark

Enterprise Value ($B)

Index Weight

American Tower AMT Towers

52.1

45%

Crown Castle International CCI Towers

40.6

35%

SBA Communications SBAC Towers

22.2

19%

114.9

100%

Verizon VZ Telecom

293.2

45%

AT&T T Telecom

242.9

38%

Sprint S Telecom

44.0

7%

CenturyLink CTL Telecom

41.6

6%

Level 3 Communications LVLT Telecom

25.3

4%

657.0

100%

Rackspace Hosting RAX Hosting

6.2

56%

Endurance International EIGI Hosting

3.3

30%

Web.com WWWW Hosting

1.3

12%

Tucows TCX Hosting

0.2

2%

11.0

100%

Clear Channel Outdoor CCO Billboards

8.2

53%

Lamar Advertising LAMR Billboards

7.3

47%

15.5

100%

Digital Realty Trust DLR Data Centers

15.7

34%

Equinix EQIX Data Centers

15.4

33%

DuPont Fabros DFT Data Centers

4.2

9%

InterXion Holding INXN Data Centers

2.4

5%

TeleCity Group TCY Data Centers

2.0

4%

CyrusOne CONE Data Centers

1.9

4%

21Vianet Group VNET Data Centers

1.8

4%

QTS Realty Trust QTS Data Centers

1.8

4%

Core-Mark Holding CORE Data Centers

1.6

3%

46.8

100%

Simon Property Group SPG Pure REIT

82.9

21%

Equity Residential EQR Pure REIT

39.9

10%

Public Storage PSA Pure REIT

38.8

10%

Health Care REIT HCN Pure REIT

37.5

10%

Ventas VTR Pure REIT

37.0

10%

Vornado Realty Trust VNO Pure REIT

33.4

8%

Prologis PLD Pure REIT

32.9

8%

Boston Properties BXP Pure REIT

32.1

8%

HCP HCP Pure REIT

31.3

8%

AvalonBay Communities AVB Pure REIT

28.9

7%

394.7

100%

Source: Capital IQ.

  • Comparing our sectors of interest yields a number of insights.  First, tower valuations are moderately correlated with valuations in adjacent or comparable sectors, particularly REITs, Telecom, and Billboards.  Reflecting towers’ blend of growth and income traits, the benchmark is equally and moderately correlated to both the S&P 500 CAPE and valuations of traditional REITs.  Second, REIT valuations have an extremely low correlation to all of the other sectors analyzed, except for towers.  As an alternative asset, REITs are thought of as a portfolio diversifier – if there were a significant positive or negative relationship to conventional assets, this benefit would be diminished.  Finally, from a portfolio construction perspective, we note that this method could be employed to “hedge” a portfolio of assets in each sector from relative upward or downward movement in multiples.  Asset pairs with correlations approaching zero are ideal.
  • With the construction of our sector indices complete, we begin to assess the long run influence of interest rates, credit spreads, macroeconomic factors, and curve slope on their comparative valuations.

Correlation of Interest Rates to Custom Sector Valuation Indices

S&P CAPE

Towers

REITs

Telecom

Hosting (1)

Billboards (1)

Data Centers (2)

2 Year

0.58

0.49

-0.19

0.57

0.19

0.41

0.38

5 Year

0.59

0.37

-0.38

0.54

0.11

0.42

0.43

10 Year

0.53

0.25

-0.50

0.45

0.08

0.39

0.45

20 Year

0.43

0.09

-0.64

0.35

0.03

0.34

0.47

30 Year

0.31

0.41

-0.34

0.23

0.04

0.36

0.54

Source: Capital IQ, Robert Shiller.  (1) Index inception 11/30/05.  (2) Index inception 1/31/2004.

  • Relative to comparable sectors, wireless towers appear to exhibit a relatively weak correlation to the absolute level of interest rates, and as noted earlier, this relationship is fairly inconsistent over the long run.  Additionally, the observed moderately positive long term relationship would suggest that a rise in interest rates may increase, rather than decrease, tower valuations.
  • While tower valuations exhibit a strong negative relationship with interest rates over the short run, this relationship has only persisted since the unprecedented events of 2008, and the direction of the benchmark’s reaction to future rate moves is unclear.  However, this correlation (like many relationships between financial data) exhibits strong mean-reverting tendencies which could indicate the inverse relationship between towers and rates has persisted for too long and that a reversion to the positive long term average is increasingly likely.

Correlation of Credit Spread to Custom Sector Valuation Indices

S&P CAPE

Towers

REITs

Telecom

Hosting (1)

Billboards (1)

Data Centers (2)

BAA Corp Spread

-0.70

-0.40

-0.28

-0.34

-0.43

-0.33

-0.29

BAA Pub Util Spread

-0.62

-0.50

-0.42

-0.29

-0.43

-0.35

-0.27

HYG – AGG Spread (3)

-0.78

-0.84

-0.16

-0.04

0.10

-0.46

-0.39

Source: Capital IQ, Robert Shiller.  (1) Index inception 11/30/05.  (2) Index inception 1/31/2004.  (3) Limited timeframe (n = 31) may bias results.

  • We also assess the effects of credit spread, or the difference in yield between AAA and lower-rated bonds of comparable maturity and duration.  As lower-rated debt issuers are forced to pay a risk premium on interest, credit spread generally reflects the market’s appetite for corporate risk, and higher levels indicate greater economic uncertainty (reflecting the ability of borrowers to service debt) while lower levels reflect increasing economic certainty and a lower risk premium.
  • Following this insight, we can see that all of the sector valuation benchmarks we have constructed exhibit inverse relationships with spread.  It is noteworthy that REITs exhibit the lowest long term relationship to spread, while that of towers does not appear to be meaningfully more correlated than that of other comparable sectors, namely Telecom, Hosting, Billboards, and Data Centers.  The outlier data point to this observation is the HYG – AGG Spread, which we use to proxy for High Yield (BB) Credit Spread in the absence of better available data.  However, the conclusions based on this data point are based on two noncontiguous periods totaling 31 observations, or less than three years of data.  As earlier noted, we are also concerned that spread may not be properly isolated as an independent variable due to the robust and persistent correlation between credit spreads and U.S. Treasury interest rates.  While correlation between spreads and valuations may be moderate, statistical significance (as represented by R2) is relatively low.

High Yield Credit Spreads and Treasuries: Strong Inverse Correlation

BB High Yield (Light Green) and BBB Low Investment Grade (Red) Credit Spreads vs. the 10 Year Treasury (Blue), 2007 to 2013, in Yield %

Screen-Shot-2016-10-12-at-11.17.48-AM

  • Following our observation of the close relationship between spread and rates, we next assess the effects of other macroeconomic indicators which may be related to sector valuations.

Correlation of Macroeconomic Factors to Custom Sector Valuation Indices

S&P CAPE

Towers

REITs

Telecom

Hosting (1)

Billboards (1)

Data Centers (2)

Macroeconomic

 

 

 

 

 

 

 

Real GDP

-0.18

0.33

0.86

-0.18

0.04

-0.07

-0.44

Unemployment

-0.73

-0.49

0.13

-0.62

-0.20

-0.38

-0.45

Market Sentiment

 

 

 

 

 

 

 

New Orders

0.03

0.45

0.89

-0.05

0.28

0.08

-0.25

Consumer Confidence

0.90

0.45

-0.05

0.73

0.37

0.46

0.45

S&P 500

 

 

 

 

 

 

 

Total Return

0.22

0.37

0.78

0.09

0.13

0.15

-0.18

CAPE

0.48

0.02

0.73

0.41

0.47

0.50

Inflation

 

 

 

 

 

 

 

CPI (YoY %)

0.37

0.39

0.14

0.17

0.35

0.18

0.25

TIPS Spread

0.68

0.49

0.09

0.38

0.53

0.42

0.45

Source: Capital IQ, Robert Shiller.  (1) Index inception 11/30/05.  (2) Index inception 1/31/2004.

  • This analysis leads to a number of insights.  Wireless towers exhibit a moderate statistical relationship to all of the factors specified, reflecting a normal cyclical relationship between economic trends and valuations.  This relationship appears to be more consistent than that of other sectors, which suggests that tower valuations are generally more cyclical than those of comparable sectors.
  • REITs exhibit markedly different relationships to pure macroeconomic factors than conventional assets: they appear unaffected by changes in Consumer Confidence but highly affected by New Manufacturing Orders.  They likewise show the lowest correlation to inflation, measured as CPI YoY % Change or the TIPS Spread, supporting the widely-held conception of real estate as an inflation hedge.  Towers’ larger and moderate correlation to inflation suggests that the sectors offers less inflation protection, but potentially more growth, than conventional real estate.

Correlation of Curve Slope to Custom Sector Valuation Indices

S&P CAPE

Towers

REITs

Telecom

Hosting (1)

Billboards (1)

Data Centers (2)

2s10s

-0.44

-0.60

-0.24

-0.51

-0.26

-0.33

-0.20

2s20s

-0.47

-0.63

-0.28

-0.52

-0.26

-0.36

-0.19

2s30s

-0.54

-0.77

-0.14

-0.66

-0.13

-0.35

-0.66

5s10s

-0.57

-0.56

-0.04

-0.61

-0.16

-0.39

-0.27

5s20s

-0.57

-0.61

-0.17

-0.58

-0.18

-0.40

-0.23

5s30s

-0.54

-0.71

0.00

-0.64

-0.03

-0.37

-0.64

Source: Capital IQ, Robert Shiller.  (1) Index inception 11/30/05.  (2) Index inception 1/31/2004.

  • The above analysis suggests that unique attributes of the wireless tower sector may result in an unusually strong relationship to curve slope, relative to both conventional REITs and comparable sectors.  While both Towers and REITs invest in fixed assets with variable but finite lease terms, the interest rate and maturity profiles of their debt tend to differ.
  • REITs tend to borrow at intermediate to long maturities (5 to 15 years) while tower companies generally borrow at short to intermediate maturities (1 to 10 years), with a weighted average maturity of debt from 5 to 6 years (reflecting that of AMT and CCI respectively).
  • We hypothesize that tower company valuations will exhibit varying sensitivity to different nodes of the curve as tower companies shift the weight of their debt across the yield curve, but only offer limited conclusions based on this insight due to the limited tenure of public wireless tower REITs.

Comparison of REIT and S&P 500 Indices

Normalized Trailing 13 Year and 3 Year Total Return of MSCI REIT (Red) and S&P 500 (Dark Green) Indices

Screen-Shot-2016-10-12-at-11.21.33-AM

Source: Capital IQ.

  • As we reorient our analysis away from correlation and toward valuation, we can see that the S&P 500 has trailed the MSCI REIT Index in total return, both over the past thirteen and past three years.  As such, we might reasonably expect a reversion to the mean of this outperformance.  We examine historical valuation relationships between the S&P 500, Wireless Towers, and Traditional REITs.
  • In particular, we will be cognizant of the potential for overvaluation in the REIT sector.  In a January 2015 report, RBC Capital Markets argued that “the REIT index is precariously perched and susceptible to a sudden shock; that shock is likely to be driven by the specter of higher interest rates as the year moves ahead.”  The Bloomberg REIT Index gained 24% in 2014, compared to an 11% gain in the S&P 500, as interest rates continued to fall after 2013’s rebound and both rising rents and occupancies boosted earnings.  Barring substantial changes in fundamentals, sectors which lead the market one year tend to lag the subsequent year.

Historical Valuation of Custom Sector Indices

Cyclically Adjusted Price / Earnings Ratio (CAPE) and EV / LQA EBITDA Valuation Multiple, Trailing 13 Year, 1 Year Moving Average

Screen-Shot-2016-10-12-at-11.24.00-AM

Source: Capital IQ, Robert Shiller, Catalyst analysis.

  • The graph above depicts the smoothed valuation of each sector benchmark in order to clarify relative trends.  Following our observation that wireless towers are both growth and income assets, tower valuations appear to be closely related to both the S&P 500 CAPE and REIT valuations.  While this provides an accurate representation of trends, it does not directly address the relative strength (over or undervaluation) of one sector benchmark with respect to another.
  • In order to accomplish this, we compare the relative valuations of pairs among the S&P 500, wireless towers, and REITs.  After taking the ratio of one sector valuation benchmark to another, we calculate the positive or negative difference of this variable to its long term average.  This method of intermarket analysis allows us to estimate how over or undervalued any given sector is to the other two.

Intermarket Analysis: Relative Value among the S&P 500 CAPE, REITs, and Wireless Towers

Trailing 13 Year and 1 Year Normalized Ratio of Select Sector Valuation Pairs, % Premium or Discount to Long Run Average

Screen-Shot-2016-10-12-at-11.26.34-AMScreen-Shot-2016-10-12-at-11.26.18-AM

Source: Capital IQ, Robert Shiller, Catalyst analysis.

  • On the basis of the above long term data, at present, towers are roughly 20% undervalued with respect to REITs and 5% undervalued with respect to the S&P 500 CAPE.  Similarly, REITs are 8% overvalued with respect to the CAPE.  These statistics lead us to a bullish view on tower valuations and a bearish view on REIT valuations.
  • While these relationships vary over the long run, relative over or undervaluation tends to persist for long periods.  In addition, each relationship can be seen to exhibit mean-reverting tendencies.  The implication of this is that significant multiple expansion may be achieved in investments in all three sectors if entered at below-average and exited at above-average relative valuations, although this describes little more than market timing.  The speed with which relative gaps in valuation among the three sectors may appear and disappear indicates that close attention should be paid to recent trends in valuation or statistically significant relationships that may affect valuation.
  • We argue that while tower stock valuations may be only moderately correlated to interest rates over the long run, both the increasing number of public tower companies converting to REITs and the nature of the negative relationship between tower valuations and interest rates after 2008 indicate that this relationship bears additional consideration, particularly in an inevitable rising interest rate scenario.

Short Term Increase in Correlation between U.S. Treasuries and Tower Valuations

Trailing 1 Year Correlation of 10 Year Treasury Yield to Weighted Average Tower Company Benchmark Valuation (EV / LQA EBITDA), Y-Axis Inverted, as of 1/30/15

Screen-Shot-2016-10-12-at-11.28.15-AM

Source: Capital IQ.

  • We observe an increasingly negative relationship between interest rates and tower company valuations over the past year, but also recognize this observation occurred in a far-from-normal interest rate environment.  It is difficult to predict the direction tower valuations would move in response to a rate change when considering the simultaneous effects of spread and slope.
  • We believe that as more tower market participants adopt the REIT structure, that the relationship between tower stocks and interest rates will remain moderately negative in the short term, before reverting to a neutral or slightly positive average following the conclusion of Quantitative Easing.
  • Both of the two strategic alternatives available to wireless tower companies present their own unique challenges.  Increasing leverage to facilitate greater acquisitions may improve growth but increase interest rate sensitivity along with the company’s reliance on cash flows to service debt.  Likewise, increasing dividend payout may improve investor income, but at the cost of both lowering liquidity (which could negatively impact credit rating, leading to an increased risk premium and higher interest payments) and repurposing earnings that could be used for accretive acquisitions toward investors.  As interest rates rise and the cost of capital increases, it may be increasingly difficult for wireless towers and REITs to lever down over time.

5) References

Bloomberg (10/14/14).  “Crown Castle Urged by Activist to Lift Dividend, Leverage.” News article.  Accessible at: http://www.bloomberg.com/news/articles/2014-10-14/crown-castle-urged-by-activist-to-lift-dividend-leverage

Bloomberg (2/12/15).  “Utilities and REITs Draw Closer as Bond Proxies.”  Chart of the day.  Accessible at: http://www.bloomberg.com/news/articles/2015-02-12/utilities-reits-draw-closer-as-bond-proxies-chart-of-the-day.

Bloomberg (1/16/15).  “U.S. Companies’ REIT Love Affairs Seen Breaking Investor Hearts.”  News article.  Accessible at: http://www.bloomberg.com/news/articles/2015-01-16/u-s-companies-reit-love-affairs-seen-breaking-investor-hearts.

Carlson, Ben (9/30/14).  “Will Rising Rates Hurt REIT Performance?”  Blog post.  Accessible at: http://awealthofcommonsense.com/will-rising-rates-hurt-reit-performance/.

Corvex Management, LP (10/14/14).  “Presentation to Crown Castle Shareholders.”  Industry Presentation and Whitepaper.

Goldman Sachs (9/2/14).  “Americas: Telecom Services Primer.”  Industry Whitepaper.

Green Street Advisors (7/1/09).  “Capital Structure in the REIT Sector.”  Industry Whitepaper.

Jacobs, John (12/5/14).  “How Should REIT Indexes Be Weighted? Enterprise Value Vs. Market Cap.  Forbes.  Accessible at: http://www.forbes.com/sites/johnjacobs/2014/12/05/how-should-reit-indexes-be-weighted-enterprise-value-vs-market-cap/.

Lazard Asset Management (12/17/13).  “Global Listed Infrastructure: Navigating a Difficult Interest Rate Environment.”  Industry Presentation and Whitepaper.

Lazard Global Listed Infrastructure Team (9/1/13).  “The Siren Song of Yield: North American Listed Infrastructure Managers Beware.”  Industry Whitepaper.

Macquarie Research (2/21/14).  “Tower and Data Center Sector Update: Historical Valuation and Interest Rate Analysis.”  Industry Whitepaper.

Macquarie Research (8/6/14).  “U.S. and LatAm Telco Services Recap.”  Industry Whitepaper.

The Deal Pipeline (8/30/12).  “REITs Demonstrate Real Appeal in More Sectors.”  News article.  Accessible at: http://www.thedeal.com/content/tmt/reits-demonstrate-real-appeal-in-more-sectors.php.

Go back to Main Research Page