Description

Coinstar (CSTR) maintains a network of about 11,000 coin counting machines in supermarkets throughout the U.S., Canada, and U.K. CSTR makes money by charging an 8.9% coin counting fee (7.5% in U.K. and for donations). CSTR’s supermarket partners benefit by receiving a portion of this fee (1% until recently), and this turns out to be one of the highest profit generators per square foot in a typical grocery store, as well as a means of converting loose change to store purchases.

This high quality, acyclical business currently trades at less than 6x EBITDA, has a positive cash balance (net of debt), and will pay minimal taxes for at least 3 years due to tax loss carryforwards. Without any major new customers or new product successes, CSTR is very likely to grow EBITDA by at least 10%/year due to increased efficiencies, new machine installs in secondary markets, and the maturation of recently installed machines (especially U.K.). Should any major new customer or product initiatives pan out, EBITDA and the stock price should go much higher.

CSTR’s service model is a natural monopoly because if there were multiple companies all doing the same thing, none of them would achieve sufficient enough geographical density to earn a reasonable return on investment. One competitive threat is that of substitutes (banks offering the service, cheap coin counters for the home, etc.), which have so far not much impacted CSTR in practice. Also a competitive threat is CSTR’s largest customers choosing to take coin counting in house like Safeway did last year.

The risk/reward for CSTR changed since it was last written up on VIC, due to the loss of major customer Safeway. The reward is lower due to higher revenue sharing with existing retail partners, but the risk is much lower because the threat of large customers leaving is much lower, as I explain below. I would strongly recommend those interested in CSTR to read through last year’s high quality CSTR writeup and followup discussion. I would also recommend listening to the last earnings CC.

LOWER REWARD

Based on hints from management in the last CC and the observation that EBITDA per machine is forecasted to be lower in 2004 than 2003 by nearly 10%, I believe that CSTR has increased its revenue sharing with its largest customers, if not all of their customers. One analyst speculates that the new revenue sharing arrangement is 1% of coins counted for the first $10 million in annual volume, and 2% for coin processing volume in excess of $10 million (the prior arrangement was 1% regardless of volume). Management has not disclosed the new arrangements so I can’t tell you what it is. But I am convinced that more revenue is being shared. While this clearly means CSTR’s profitability going forward is less than what one would have expected a year ago, it also means that customers have less of an incentive to drop CSTR in favor of starting their own coin counting program like Safeway.

To reflect pessimistic assumptions on revenue sharing going forward, I project for 2008:

Machine Count: 14,600
Revenue:$275 million
EBITDA: $88 million
Income: $34.3 million
Shares: 14.4 million (assumes all free cash flow used to buy back shares)
EPS: $2.38 (fully taxed)

This scenario assumes no major new customers, that all new products/services fail, and that CSTR drops CapEx to maintenance + new machine levels starting in 2005 (If all the new products and services are bombs, then presumably they will stop investing in them). Yet, at 6x EBITDA (and P/E of 15.5), the stock price would be nearly $37 per share in 2008, an annualized return of 25% from today’s price. Obviously CSTR could do better if any of the major new product or new customer efforts pan out.

As for downside, consider that CSTR’s market capitalization is about what it would cost to build CSTR from scratch, and it is not difficult to imagine a few potential acquirers for whom buying CSTR would be a nice complement to their existing product/service lineup. During the point of maximum pessimism after CSTR lost Safeway, when many were worried they might lose Kroger as well, the stock price dipped temporarily below 12.

LOWER RISK

Given that Safeway terminated its relationship with CSTR last year and decided to take coin counting in house, the most obvious question to be answered is whether Safeway will succeed and therefore be imitated by CSTR’s other two large customers Kroger and Albertson’s.

Anyone who listened to the last earnings CC would have to conclude that Safeway is off to a terrible start with their coin counting project, as 33% average down time is not only leading to customer frustration but also to unusually high comps for Coinstar machines located near Safeway stores. So far, Safeway’s initial difficulties are an advertisement for CSTR’s services. We know from Coinstar’s history that this business model requires a fairly high density of machine locations and logistical coordination in order to function smoothly and profitably.

Specifically:

* CSTR has a number of patents on coin counting machines. Their main competitor ScanCoin is so far choosing to honor the patents, leading to machines which break down more often and have more difficulty dealing with noncoin debris. One would therefore expect higher service costs per unit on competitor machines.
* CSTR machines are networked to a central computer that optimizes service call frequency and responsiveness, as well as coin transport efficiency. It also greatly reduces the possibilities for fraud. This system infrastructure took many years and $10s of millions to develop.
* Technicians must travel between stores; a single chain of stores has less density than the multiple chains serviced by CSTR. A less dense network must have more technicians per machine to cover the increased travel time.
* CSTR did not turn EBITDA positive until it had over 5000 units installed. Though they could have turned EBITDA positive with fewer units if they were all located in the same metropolitan statistical area (i.e. U.K. was EBITDA positive with 500 machines), it is doubtful any existing chain has sufficient location density to make more money at this business then they could by using CSTR’s services.
* Most CSTR customers have experimented with competitor coin counting machines in the past but have concluded that sticking with CSTR would be more economical than servicing coin counting machines on their own. Safeway is the anomaly and so far they are off to a poor start.

All that being said, it may be possible for Safeway to succeed at the coin counting business in the long run do to unforeseen breakthroughs in technology or logistics. But it is certainly clear from their rocky start that it will take significant investment and management attention to make it work, and is therefore a distraction from their core business. Coinstar’s three largest customers Kroger (22.3%), Alberston’s (11.8%), and Ahold (approximately 8%) are clearly not intending to follow Safeway in the short run as they all recently signed multi-year contracts and have agreed to increase the number of locations that have Coinstar units (though it should be noted that the latest 10K indicates that some customers can cancel with 6 months notice). In fact, I waited for the last of these three contracts to be announced (3/11/04) before investing. Grocers have a tendency to mindlessly imitate their competitors, but they aren’t imitating Safeway in this case and they are unlikely to so long as Safeway continues to struggle with its program.

Of particular note is the resilience of CSTR in spite of losing Safeway. 2003 results were within the initial guidance range laid out 11/20/02 in spite of the lost Safeway business. The company engaged an independent firm to monitor’s Safeway’s (so far) disastrous launch and is using this information to promote CSTR’s superior technology and servicing abilities. CSTR is refurbishing all of the de-installed Safeway machines to like-new condition for use in other stores, which reduces capital expenditures for new machines. Not bad after losing a 10% customer. The fact that CSTR has done so well in spite of losing a 10% customer actually increases my confidence in this company.

SAME MACHINE GROWTH

There are three reasons to believe that mid single digit same machine growth will continue for at least the next few years.

The most obvious reason is that less than 20% of the population within a few miles of a Coinstar machine have actually tried it. The company estimates that 80% of those who try the service once become repeat customers, and that there are 16,000 to 18,000 viable locations for Coinstar units in their 3 nation service area. If CSTR’s near monopoly status holds, I think CSTR easily has the potential to process over $4 billion in coins per year in the U.K., Canada, and U.S. (.25 million per unit, 16,000+ units). This is double the $2 billion processed in 2003. Management claims that increasing “trial” is their highest priority because it is their highest ROI opportunity.

A second reason to expect continued SSS growth is that the new backlit models cause an immediate increase in sales of about 10%, according to the company. Older machines will be gradually upgraded to the backlit model over the next few years.

A less obvious reason relies on the historic growth trajectory of an average Coinstar unit. The company does not disclose revenue or coins processed per machine by age category. Using a number of indirect pieces of information given out by press releases, SEC filings, and conference calls, I believe the aging to be approximately as follows for total dollars processed per machine per year (I kept changing these numbers until known information was close to correctly modeled):

Year 1: $48,000 (assumes average machine installed on July 1)
Year 2: $125,000
Year 3: $150,000
Year 4: $170,000
Year 5: $187,000
Year 6: $203,000
Year 7: $218,000
Year 8: $229,000
Year 9: $236,000
Year 10: $240,000

Using these assumptions and publicly available information such as machine counts and Safeway data, I was able to closely model CSTR’s operating statistics (including Safeway’s percentage of sales) for all years since 2000 except 2003 (which was thrown off by Safeway’s deinstallations occurring towards the end of the year as opposed to July 1 on average).

At any rate, this model implies over $20,000 revenue per year for Coinstar units over 7 years old, a considerable increase from the 2003 level of $16,048 per machine. As CSTR approaches saturation 5-10 years from now, I expect the annual revenue per machine to exceed $20,000. This does not even factor in the benefits from backlit displays and increased exposure to a greater percentage of the population. I can supply more detail about my model in the discussion followup if anyone is interested.

UNITED KINGDOM

Rarely remarked upon by the investor community is CSTR’s success in the United Kingdom with the #2 and #3 chains (Sainsbury and ASDA). After several years of testing a small number of units, CSTR began installing large numbers of units in 2001. Three years later, its United Kingdom network is more profitable than its North American network, in spite of the lower 7.5% fee. The company stopped supplying detailed operational data breakdowns between North America and U.K. a year ago (Perhaps they don’t want to make it obvious to their U.K. customers that they are so profitable as they might be tempted to go the route of Safeway), so it’s getting harder to tell what is happening. However, I believe the following:

650 to 700 units in U.K. as of year end 2003

Per UK unit revenue (converted to U.S. $) is between 75% and 100% higher than North American units in similar age categories (I constructed a similar unit-by-age-category model to fit the known data).

I am not expecting much more unit growth in the U.K. until CSTR signs on the #1 or #4 chains (Tesco & Morrison’s/Safeway). This is because Sainsbury and ASDA combined have less than 800 stores. It is likely that some of the 800 stores are in locations that can’t be economic for CSTR, so there is probably little unit growth remaining at these two chains.

Even without any new units, CSTR’s U.K. revenue will likely exceed $15 million in 2004 based on machine aging and the more favorable currency translations. Net income in 2004 should exceed $6 million from U.K., which represents over 30% of CSTR’s net income. Should the dollar continue falling, this percentage would obviously be higher.

If Tesco or Morrison’s became a major Coinstar customer, U.K. net income could jump an additional $5 million or so.

NEW PRODUCTS

CSTR’s web site has details about new product initiatives in archived press releases. Frankly, I’m somewhat skeptical as to whether they will succeed in the highly competitive prepaid wireless and MasterCard arenas. Not only are these products available by a variety of distribution channels, but there is also direct competition from Gtech, who is rolling out these capabilities to lottery machines already installed in several convenience store chains. And I’m puzzled by the latest acquisition of CellCards of Illinois, which brokers these types of services for distribution at check-out lanes. But based on past history, I do trust that management will bail out of all these prepaid initiatives if it becomes obvious they won’t produce adequate returns on investment.

I do very much like the risk/reward on the payroll card partnership with FSV. CSTR’s investment and marketing support is immaterial, while they get a small fee from FSV every time a user prints statements off of a Coinstar unit. Coinstar has nothing to lose and possibly a lot to gain. I hope this indicates a future direction for Coinstar; leveraging their existing network of machines to provide services at minimal cost. All it takes is software updates on the central system to add such capabilities to the entire network.

The overall theme to Coinstar’s new product initiatives seems to be making available bank-like products to the unbanked. It is interesting to note that Wal-Mart estimates 20% of Wal-Mart’s customers do not have bank accounts. Legislation currently winding its way through congress would effectively bar Wal-Mart from opening bank branches in their stores. So that increases the chances that Wal-Mart might pick up some of Coinstar’s product offerings that serve the unbanked.

OTHER COMMENTS

Coinstar management speaks intelligently about capital allocation and has a good track record to back it up. They terminate projects unlikely to work out (i.e. meals.com). They buy back stock when the stock price gets low. They limit the risk of new products or new markets by testing small numbers of units before larger rollouts. And their acquisitions have been out-of-favor companies at a cheap enough price that the risk is minimal, while having significant upside potential.

Coinstar management has historically issued conservative guidance, and I believe 2004 guidance continues the tradition. Based on the coinstar unit revenue by age model I described above, I believe that revenue for 2004 will almost certainly hit the high end of guidance, if not higher. Income is harder to tell, as the combination of more revenue sharing and cost cutting initiatives is something we can’t well model until we’ve seen another 2 or 3 quarters of results. But I wouldn’t be surprised if management was conservative on the income as well, based on past trends.

Q1 will not be a great quarter for CSTR as it is the first quarter fully without Safeway. Revenue for the first 12 months of a newly installed machine is typically 40% to 50% of a mature machine, according to management. I believe that most of the replaced Safeway machines were installed in 1998 and 1999, so each Safeway unit replaced is at least a 50% per unit revenue hit. CSTR was also slightly impacted by the large grocery strike that just ended.

CSTR’s stock price has been dropping ever since Suntrust put out a negative research note on Coinstar earlier this month. I saw this report and was not impressed. The analyst rehashed the same argument that large grocers (Kroger, Albertson’s) can do coin counting on their own as a high return investment. The backup analysis to support this contention was quite meager (not taking into account greater technician travel time, lower reliability of ScanCoin units, and the several year learning curve to the service side of this business). There was also no mention of facts such as Safeway’s difficult start, and that many chains have experimented with ScanCoin and chose to stick with CSTR. Nevertheless, the stock price has been dropping on higher than normal volume ever since this report was released.

What I like least about CSTR is that there has been substantial management turnover during the past few years and there has been very little insider buying, even when the stock price dropped to its lowest points. I like to see substantial insider ownership - not the case here.

RISKS

Safeway succeeds – This would have a much bigger impact than any other risk I can think of, but I see it as a low probability event. If they do succeed, expect Kroger and Albertson’s to take coin counting in house. If they don’t succeed, then expect them to return as a CSTR customer.

Gradual margin erosion as largest customers demand lower fees each time contracts are renewed.

Turmoil among grocer customer base: strikes, Wal-Mart competition, mergers, etc.

Substitutes always a risk.

One or more large banks decide to follow the lead of Commerce bank, rolling out coin counting machines at all bank branches. This is very expensive for banks so I see this as unlikely.

Technology advances drastically lower cost or increase reliability of competing models of institutional coin counting machines. The machines would probably then become a staple at banks.

Catalyst

CSTR raises guidance for 2004
One of the new product launches begins to succeed in a material way
Major new Coinstar unit deployments (such as Publix in U.S. or Tesco in U.K.)
Wal-Mart installs some of CSTR’s new services targeting the “unbanked”
Safeway returns as a customer
CSTR acquired

Many of these catalysts individually are unlikely, but taken together it seems likely that one of these good things is likely to occur within a year or two.