Description

Company is located in Indonesia, but I chose Malaysia because it wasn't an option

Investment thesis

• Sarana Menara is the largest independent tower operator in Indonesia under the brand Protelindo with 16,790 towers (average age of 7 years and built to handle 4 or more tenants) and 27,918 leases in its portfolio. It was founded in 2008 and quickly grew to become the largest player through some shrewd acquisitions
• Share price has fallen 40%+ over the past year because of 3 main reasons: i) Largest customer (36% of FY17 sales) has leases coming up for renewal from 2020-2023, ii) Telecoms in Indonesia aren't doing well and fear of pricing pressure + mistakenly grouped together by investors, and iii) Weak overall macro sentiment with Rupiah depreciation and "crisis" in ASEAN EMs
• I believe the market has turned overly pessimistic and is pricing a near 100% probability of non-renewal from its key customer (Hutch, 4th largest telcom). Additionally, the company is trading below replacement cost of 8x EBITDA and it is the cheapest towerco domestically even though it is the only one with a strong presence in the faster growing microcell pole market through its acquisition of iForte and telcos' price war has ended and are starting their capex again.
• Management style: Controlling shareholder is Hartano family (richest in Indonesia) who has 50.1% and quite a good reputation. When it comes to running the company, they are very hands-off and prefer to hire professionals. Ex-CEO of Crown Castle (2nd largest towerco) is an adviser through PE investor Berkshire Partners (not to be confused with Berkshire Hathaway). Also open to listening to investors as they have announced a share buyback program of up to 5% of CSO after discussions with shareholders

What is a tower and how do they make money?

• In a very general sense, a tower is a vertical structure built on a piece of land designed to accommodate multiple tenants (telcos) - macro towers are 100+ ft while height of micro towers are below 100 ft
• Tenants lease vertical space on the tower to place their communications equipment (antenna, microwave and cables). In a very general sense, you can view towercos as landlords with towers being their "buildings" and telcos are their tenants
• Contract length and fee: Leases in Indonesia are typically 10-year leases and Protelindo charges Rp 13m per month with telcos pre-paying a year in advance (same price regardless of anchor tenant or colocation for Protelindo)

What are Protelindo's main expenses and why are margins so high?

Capex and D&A

• Overall, Protelindo spends about Rp 1.3 trillion (24% of FY17 sales) on capex
• Initial capex: On average each macro tower costs Rp 900m to build from scratch (EBITDA payback period of 7.5 years at 80% margin) and Rp 500m for a micro tower but the lease rates are similar. Important to note that Protelindo will only take the contract if the tower site meets its minimum ROIC criteria and does not build a site without securing a site lease with an anchor tenant. Main capex are:
  • Land acquisition cost, estimated to be 20-25% for macro tower and 40% for micro tower. Protelindo usually pre-pays a 10-year lease and start negotiating an extension in year 7 or 8 with the government
  • Material (mostly steel), clearing the land and construction costs make up the rest
• Colocation capex: Once the structure is built, capex for any subsequent tenant (termed as colocation) would be ~Rp 100m (~US$6k) for Protelindo, meaning payback period of 10 months and 95%+ EBITDA margins for the next 9 years and 2 months
• Acquisitions: Protelindo has a very good track record of successful bolt-on acquisitions, spending Rp 3.2 trillion (~12% of current market cap) since 2012 to acquire ~5500 towers (33% of current towers) with average consideration of 8x EV/EBITDA
• D&A: 17% of FY17's revenue but lower than average D&A expense since IPO of 23% because of a change in useful life estimate from 20 years to 30 years in 2017. Nevertheless, since all the cash is paid upfront and very little to replacement cost, an argument can be made that we could add back at least 50% of D&A to see Protelindo's operating owner earnings because majority of the non-cash expenses won't need to be replaced (I estimate the biggest "replacement" capex will be renewals of land lease which is 25% of initial capex)
  • Data point: American Tower (largest towerco in the world) spent 12% of sales on capex with only 2.6% going to maintenance on average since 2008

Cash expenses

• ~71% of their costs are fixed, giving them high operating leverage when a tower has multiple tenants
• Cost breakdown:
  • Cost of revenues (5.2% of sales) - Mostly site maintenance (security guards and cutting the grass), electricity and equipment rental
  • Selling and Marketing (1.5% of sales) - Salary of salespeople and travel expenses
  • General and administration (7.1% of sales) - Salary of technical and admin staff, consultants, permit costs and office supplies
• Core EBITDA margins (excluding other income/expenses) have been consistently between 82.9% to 87.2% from 2008-2017, and according to management the tower business has 90%+ EBITDA margin in recent years due to their economies of scale

Returns - Towerco economics and why telcos aren't doing it themselves

* ROCE calculated as EBIT/Capex, actual ROCE will be higher because Protelindo has negative NWC as telcos typically prepay for the entire year

• Important to understand that with only 1 tenant, the returns aren't very attractive, hence scale and tenancy ratio (number of leases / number of towers) are crucial to succeed in the industry
• A key reason that telcos are trying to outsource their tower business is because it is very capital intensive and the returns are low for only 1 tenant (telcos are unlikely to allow competitors to have a colocation on their tower). Telcos have been gradually deleveraging and trying to become more capital-light, thus Protelindo acquired the tower portfolios of XL Axiata and Hutch while Tower Bersama (TBIG) bought Indosat's towers

Industry - Growth drivers and competitive dynamics

Headwinds since 2015

• 2015 to H1 of 2018 was a tough period for the telecom industry due to the fierce price war between the 3 main telcos - Telkomsel, XL Axiata and Indosat. The price war gave customers falling data prices, leading to an explosion in data usage
• Background of the price war: Pre-paid SIMs are 98% of the market because less than 20% of the population even have bank accounts. The telcos started offering better and better initial bonus, leading to a very high churn rate in SIM cards as users will continually change when the initial bonus finishes
• As a result, both XL Axiata and Indosat postponed expansion plans as revenues were either flat or posted low single digit growth, data traffic on all the telcos networks have at least tripled and traffic per BTS (Built to suit tower) has also risen dramatically. This is especially obvious for Indosat who has been lagging behind the other 2 in terms of network coverage

What has changed?

• In May 2018, the government regulated that SIM cards had to be registered to lower the churn rate and to stop the price war. Telkomsel announced that they will start raising prices in H2 2018 (around September), XL Axiata followed soon after and Indosat will probably follow suit to improve profitability. Telkomsel, the market leader, is under pressure from shareholders to become more profitable instead of simply prioritizing revenue growth

• Around the world - what is the saturation?
  Country	Number of towers	Population (000s)	Ratio (000s)
  United States	215,000	361,000	1.68
  China	1,840,000	1,379,000	0.75
  India	459,395	1,324,000	2.88
  Vietnam	70,000	92,700	1.32
  Thailand	52,483	68,860	1.31
  Philippines	16,600	103,300	6.22
  Malaysia	22,682	31,190	1.38
  Myanmar	14,303	52,890	3.70
  Japan	220,000	127,000	0.58
  Bangladesh	30,000	163,000	5.43
  Average	294,046	370,294	1.26
  Indonesia	90,000	261,000	2.90

  To get an "outside view", I compared the penetration of towers (measured by population per tower) in a mixture of developed and developing countries to understand what could be the upper limit for the country. There seems to be a strong correlation between a country's development and the penetration. The countries with lower penetration than Indonesia (those highlighted) would generally be classified as less developed than Indonesia
• Protelindo estimates saturation would be ~200,000 towers and based on the table above, the figure is in-line with our primary research
• Indosat and Hutch have both won more spectrum last year in the auction for US$31m each. Estimated that they would only be done with spectrum refarming in April, but as of June 2018 they have not started network expansion yet.
• The second driver for Protelindo is the densification of towers as telcos move from 2G to 3G and 4G nationwide and better devices (5G will in 2022 at the earliest)
  • As a rule of thumb, lower frequency spectrum provides a larger coverage area while higher frequency covers shorter distances but is faster. So when spectrum usage increases, the distance it can propagate decreases.
  • Furthermore, when devices become more advanced, the demand for higher bandwidth applications and higher service quality result in a narrower range at which signals can be transmitted. This is leading to a demand for denser networks for carriers, called densification
• Thirdly, Protelindo's management had sufficient foresight (or open to listening to outside advice) to acquire iForte in 2015, making them the only independent towerco with a big presence in the microcell towers and fiberisation of towers. In Jakarta, where land is scarce, microcell towers are the only way for telcos to increase the density of their networks. Due to the growing need of non-macro tower telecom infrastructure, its contribution has grown from 3% to ~10% in 3 years time. Management thinks this segment's contribution could reach 15-20% in the future
  • Total addressable market: Target markets will be developed cities, Java island has 6 of them - Jakarta will require at least 2,000 and the other 5 will need 1,000 each
  • Margins will be lower than macro towers, but will still be above 70% EBITDA margin and have higher ROCE because capex for these microcell towers are 40-50% of the macro towers and will charge ~Rp 9m per month vs Rp 13m for macro towers
• In summary, Protelindo is on track to having its best year for organic growth since 2013, commencing 1309 leases in H1 2018 vs 1591 for FY2017

Competitive landscape

• Excluding towers owned by telcos and their subsidiaries, there are ~42,000 towers in Indonesia. Protelindo recently acquired the 4th player (KIN) who had 1,369 towers after talks to acquire STP fell through (might be possible in the future as Carlyle is a large shareholder of STP and is looking to cash out after 6 years)
• In the past, there were 100+ towercos in the fragmented market - barriers to entry were low and all you needed was some land and steel to build a tower. The market was inefficient and the only companies that had a large portfolio of towers were telcos
• With the emergence of "professionally" run towercos like Protelindo and Tower Bersama (TBIG), the market has started consolidating. For example, Indosat used to have 47 towercos serving it a few years ago, now it only works with 27 as the smaller guys who don't have scale are getting acquired. Combined, the 2 players have 72.8% of the independent tower market
• To a telco, there are 3 main criteria when choosing a towerco: i) Ability to negotiate for the new lease rate, ii) Delivery speed that gives the telco coverage in that area to compete against other operators, and iii) Accessibility to the site
• When measured up against these 3 criteria, it is apparent that scale has become a huge competitive advantage for Protelindo and TBIG while STP has not increased tower count since 2015. The 2 big players can guarantee fast delivery either buy building new towers (each has capacity and financial means to build new towers) or simply let telcos colocate if they have an existing tower in the area (very appealing for telcos who want to work with towercos that can give them a portfolio of hundreds of towers rapidly)
  • Colocation is the most bang for buck as it takes 4-5 weeks while building a new macro tower typically requires 6-9 months
  • When I asked TBIG about competition from Protelindo in March 2018, their response was - "No competition as both companies try not to build towers beside other towers, telcos either use their space or don't enter that region"
  • Follow up call with TBIG in Sep 2018 on locations - "Don't compete on colocation because we have unique tower locations"

The elephant in the room - Overhang of Hutch renewal

What is the current situation?

• Going back to the inception of the company, the Protelindo we know today got its first start by acquiring ~3600 towers from Hutch in 2008
• Over the years, revenue contribution from Hutch decreased from 50% in FY08 to 36% in FY17
• When Hutch was given the choice in 2008, they opted to pay in USD, a choice which has benefited Protelindo greatly as Rupiah has depreciated almost 40% against the dollar
• Based on the current contract of ~8000 leases, 40% of Hutch's leases are up for renewal in 2020, and 20% thereafter for each year from 2021-2023
• In 2017, Hutch's ex-CEO was rumoured (no news articles and CLSA doesn't know anything about this) to be playing hardball and threatened to not renew any lease with Protelindo - he was replaced shortly after and the new CEO has been very open to discussions with Protelindo

What are the possible outcomes?

• Worst case - Lose all Hutch revenue: I believe this is very unlikely as i) TBIG said it has little overlap with Protelindo so if they want Hutch's 8000 leases they have to start construction now (TBIG said it would be impossible), ii) Typically takes a few months for a telco to relocate equipment from one tower to another. During that period, the telco will lose network coverage (assuming it doesn't buy a new set of equipment - TBIG estimates each set will be US$25k) and might lose customers as low switching costs due to 98% using pre-paid SIMs. Once they gain notoriety of having "bad service", hard to gain customers back, iii) Hutch just bought more spectrum in the last auction (October 2017) so it seems unlikely that they will exit the market
• Best case - Retain all Hutch revenue: Even more improbable because Hutch is paying almost Rp 20m per month (compared to current market price of Rp 13m) because they're paying in USD rates from 2008
• Expected outcome - Renegotiate lease prices and lose some leases: Most likely scenario is that Hutch decides to pay in Rupiah at the current market rate which would mean a 35% decrease in lease rate. Additionally, I expect Protelindo to lose some of its 8000 leases to competitors or reduced coverage by Hutch

Does TBIG think they can steal Hutch from Protelindo?

• Based on a call with TBIG in September, they said towers are not widgets, Hutch cannot simply walk away from Protelindo as there is very high switching costs to send men to take down 8,000 boxes across the country and ensure the network is operational. Management thinks that if that really happens, Hutch is finished
• As long as Hutch remains in Indonesia, they will probably continue but negotiate to pay in Rupiah instead of USD this time

Valuation

Expected outcome: 76% upside

• Expect 6% revenue growth (management guided 7%+) and 60% EBIT margin for the next 2 years, Hutch revenue flat so non-Hutch growth is 10% (consistent with the previous years)
• From 2020 to 2022, Hutch revenue will decrease by 20% annually, ending with 40% of FY17's Hutch revenue in FY22 and EBIT margin decreases to 55% due to negative operating leverage (operating costs are Rp 2264bn in FY18 vs Rp 2595bn in FY22)
• Breakdown of Hutch revenue lost: 35% from USD to Rupiah (20m to 13m) and ~40% of leases are not renewed (8000 to 4800)
• EBIT margin recovers to 65% in terminal state (80% EBITDA margin and 15% D&A and Capex because American Tower's maintenance capex is 2.6% or 1/6 of current capex spend)

Worst case outcome: 3% upside

• Hutch cancels all leases as they expire losing Rp 1.9 trillion of sales by FY22, EBIT margin decreases to 50% from FY20 till TV
• Terminal Value assumptions: EBIT margin increases to 55% as D&A decreases from 20% to 15% in the steady state (reason explained above)

Peers around the world

Ownership, management and advisers

Ownership and Management

• Owner: Owned by the Hartano family (richest in Indonesia 9 years consecutively, estimated that 1 in 5 cigarettes in the country is from them) - have been increasing stake from 32% in Sept 2017 to 50% in August 2018
• Management: Run very professionally, family doesn't meddle too much in operations as they prefer to hire experienced guys. CEO was business development head for the family for 20 years, CFO spent 30 years in various positions in commercial and investment banking, CMO has been in tower business for over 15 years and Chief Business Support (HR, IT and Treasury) worked 15 years with the family
• Compensation: Senior management are given targets on EBITDA, leases and net profit by the board. They only get a performance bonus if they meet all 3 KPIs, 30% of the bonus will be in shares
• Other prominent shareholders: T. Rowe Price (8%), Overlook (5%), Harris Associates (1%), Putnam Associates (0.18%) and Grandeur Peak (0.12%)

Advisers

• Through Berkshire Partners, the fund has a senior adviser named John Kelly who has been in regular contact and giving them very useful advice. For example, Protelindo learned how to better serve their customers and the evolution of renewals in the US to improve business processes in Indonesia
  • Background: John Kelly, CEO of Crown Castle from Aug 2001 to Jul 2008, during his tenure the share price appreciated from ~$8 to ~$36 in 7 years. After he stepped down, he remained on the board until 2016 when the shares averaged $85

Risks

• Biggest risk is that Hutch quits the industry - that is the only scenario where Protelindo doesn't get any more business from Hutch. This is not an impossible event because the market is very competitive
• Telco consolidation - will hurt all towercos as there will be less colocations, decreasing the tenancy ratio across the board. Unlikely to happen as a business decision but could always be a political play by the government
• Another price war among telcos
• Currency depreciation - non-fundamental but might spook international investors from looking at Indonesian equity markets
• Change in regulation to allow foreign firms to own towers - currently foreign firms aren't allowed to own towers
• 25% of debt is in USD which was to hedge revenues from Hutch. Management told me that they plan to do a refinance to Rupiah if Hutch decides to pay in Rupiah after their contract negotiation.
• Quite illiquid and high foreign ownership - share price might be weak as long as foreign outflows continue. Buyback could exacerbate illiquidity as well especially since free float is quite low

Catalyst

1. Announcement that Hutch is renewing their leases will lead to an immediate re-rating on the stock. It has historically been trading at a premium to Tower Bersama and only started trading at a discount this year

2. Acquisition of STP - Carlyle is the largest shareholder and it is an open secret that they are looking to exit. If the market continues to deteriorate, I think that Protelindo will be able to purchase them at a bargain